Microsoft.VisualStudio.Copilot

An object that represents a code vulnerability annotation.

The type of vulnerability. The name of the type to display, as provided by the server. Used as a fallback if we don't have client-side strings for the type. The description of the type to display, as provided by the server. Used as a fallback if we don't have client-side strings for the type.

An object that represents a code vulnerability annotation.

The type of vulnerability.

The name of the type to display, as provided by the server. Used as a fallback if we don't have client-side strings for the type.

The description of the type to display, as provided by the server. Used as a fallback if we don't have client-side strings for the type.

An enum representing the different types of code vulnerabilities that can be detected by Copilot.

An unknown vulnerability type. Used as the default when the server returns a type that is not recognized by the client.

Unvalidated input in path value creation risks unintended file/directory access.

Unchecked input in database commands can alter intended queries.

Treating external input as code without proper validation.

Allowing user input in system calls without proper escaping can allow arbitrary command execution.

Error messages or stack traces can reveal sensitive details.

Unvalidated input in URLs risks syntax or behavior modification.

Debug mode in Flask production allows arbitrary code execution.

Logging sensitive info like passwords by exposing it in clear text.

Regular expression may match unexpected hostnames.

Unvalidated URL redirection from a remote source.

Allowing unvalidated redirection based on user-specified URLs.

Unvalidated URL redirection from a remote source. TODO: differentiate from ServerSideUnvalidatedUrlRedirection.

Using weak encryption or hashing algorithms poses security risks.

Using weak encryption or hashing algorithms poses security risks. TODO: Differentiate from WeakCryptographicAlgorithm.

Unvalidated input in HTTP requests may lead to SSRF (Server-Side Request Forgery) attacks.

Binding to all network interfaces may expose traffic interception.

Adding unvalidated user input to HTTP responses invites XSS.

Storing sensitive data like passwords in clear text is insecure.

Storing sensitive data like passwords in clear text is insecure. TODO: Differentiate from ClearTextStorageFile.

Extracting tar archives without validation may lead to files being unexpectedly overwritten.

Embedding credentials in source code risks unauthorized access.

Weak random number generation can compromise security.

Weak hashing algorithms for passwords poses security risks.

Permissions for files should be kept as strict as possible to prevent unintended access.

Unvalidated user input used in regular expressions can lead to arbitrary code execution.

Incorrect usage of the sizeof operator on a pointer can lead to memory corruption or buffer overflow vulnerabilities.

Using a variable that may not have been initialized can lead to unpredictable behavior or crashes.

Represents the type of error associated with a response.

Catch-all error type. This will display a generic error message and direct the user to the output logs.

An object that represents a chat agent error annotation.

The type of error. The message of the agent error.

An object that represents a chat agent error annotation.

The type of error. The message of the agent error.

The type of error.

The message of the agent error.

An object that represents an annotation for cited intellectual property.

The snippet of the code that is cited. The URL to the code snippet. The type of intellectual property that is cited. The license for the cited intellectual property.

An object that represents an annotation for cited intellectual property.

The snippet of the code that is cited. The URL to the code snippet. The type of intellectual property that is cited. The license for the cited intellectual property.

The snippet of the code that is cited.

The URL to the code snippet.

The type of intellectual property that is cited.

The license for the cited intellectual property.

A base implementation of progress that maintains order call to an async implementation of Report.

The type of progress to report. As this is not JTF aware, creation and calls to must remain in the same async flow or risk deadlocks.

The latest version of the Copilot service.

A list of all historical responder versions in descending order. Used to locate responders that are compatible with a given version.

Provides functionality for encoding text into tokens and counting tokens, optimized for use with Copilot language models.

This is an exact copy of in terms of encoding, however when counting tokens it will use a faster approach that is not exact but gives a good approximation.

The function trimmed a piece of text by suffix based on token count needs to be preserved

Text to be trimmed by suffix based on token count Max token count to be preserved Trimmed string

The function trimmed a piece of text by prefix based on token count needs to be preserved

Text to be trimmed by prefix based on token count Max token count to be preserved Trimmed string

Retrieves an instance of an ICopilotTokenCounter.

approximate counter; otherwise, to use a precise counter. An instance of . Returns a fast, approximate counter if UseApproximateTokenCounter is ; otherwise, returns a precise counter. Prefer using approximate counter when working in-proc.

Sets the denominator for the fast token counter.

The denominator value to use for token counting approximation. This method is intended for testing purposes. An internal ResetForTesting method was initially considered, but due to InternalsVisibleTo complications with multi-targeted test projects (.NET 8 and .NET Framework 4.7.2) referencing a .NET Standard 2.0 library, a public method approach was chosen instead. Use first to enable approximate counting, then use this method to adjust the denominator between tests.

Resets the internal state used to track the approximate token count for subsequent operations.

This method is intended for testing purposes. An internal ResetForTesting method was initially considered, but due to InternalsVisibleTo complications with multi-targeted test projects (.NET 8 and .NET Framework 4.7.2) referencing a .NET Standard 2.0 library, a public method approach was chosen instead.

A formatter that is hard-coded to use the for values in a dictionary.

Determines if a parameter accepts null values (either nullable value type or nullable reference type).

This options class is used to override LoadType to ensure type loads occur on the AssemblyLoadContext that loaded Copilot assemblies. In case of ServiceHub and extensibility scenarios, this will be extensions load context which will be able to resolve assemblies deployed by the extension.

A class that serves as a collection instances to handle multiple tracing strategies.

A class that holds as a collection of objects that can be disposed when the object is disposed.

A class that represents a tracing strategy that uses the object set in . If is not set or does not implement , then this strategy does nothing.

Represents a language model configured for use with Bring Your Own Key (BYOK)

Gets or sets the API key used to authenticate requests to the external service.

Gets the name of the model provider.

Gets the maximum number of output tokens that the model can generate in a single response.

Initializes a new instance of the class.

The raw identifier of the model. The name of the model, such as gpt-3.5-turbo-0613. The display name of the model, such as "GPT 3.5 Turbo". The model family name, such as gpt-3.5-turbo. Name of the model provider. The number of tokens that can be provided for context (request, context, history). The API key used to authenticate requests to the external service. The capabilities of the model. The maximum number of output tokens that the model can generate in a single response.

Represents a custom bYOK that has a custom URL endpoint.

Gets the model Id.

Gets or sets the displayname of the model.

Gets or sets a value indicating whether the model supports tool calling capabilities.

Gets or sets a value indicating whether the model supports vision capabilities.

Gets or sets maximum number of input tokens supported by the model.

Gets or sets maximum number of output tokens supported by the model.

Gets the provider Name.

Gets a value indicating whether the model is custom.

Gets or sets a value indicating whether the model is selected.

Gets or sets the custom URL endpoint for the model.

Gets or sets the name of the model provider.

Gets or sets a value indicating whether the API key is available for this provider.

Gets or sets the models for this provider.

Gets or sets the Endpoint for this provider.

Stores default configuration settings for BYOK models.

Default maximum number of input tokens supported by the model.

Default maximum number of output tokens supported by the model.

Default value indicating whether the model supports tool calling capabilities.

Default value indicating whether the model supports vision capabilities.

Represents a command invocation action.

Initializes a new instance of the class.

The name of the command to invoke. This name should be unique within the service represent by . The service moniker of the service which provides the command handler.

Gets the name of the command to invoke.

Gets the service moniker of the service which provides the command handler.

Gets the arguments for this command.

Options used during provide command for .

Gets a value indicating whether the provided command can work with the default commands.

Defines a user provided command.

An event to notify consumers that the value of might have changed. Consumers should call after receiving this event to check the current value.

Executes the provided command.

Due to the asynchronous nature of the method, a 'false' return value from does not guarantee that this method will not be called. Therefore, implementations should ensure internal validation.

Determines whether the provided command can execute.

Convert the content of this interaction to Markdown, removing and replacing mentions.

The interaction to convert. A markdown formatted string of . This promptifies the content of the interaction, removing agent and command mentions and replacing context mentions with links to the context.

Convert this content to Markdown, removing and replacing mentions of from the .

The interaction to check mentions of context against. A markdown formatted string of the content. This promptifies this content, removing agent and command mentions and replacing context mentions with links to the context.

Convert this content to Markdown, removing and replacing mentions of from the .

Converts the mention string into a uri for markdown.

The mention to convert. A markdown uri of the mention.

Strongly typed identifier for a copilot action (confirmation, function, etc).

Initializes a new instance of the struct.

Unique for the action.

Gets the unique identifier for the action.

This should be the matchable to the identifier that the endpoint provides when indicating an action.

Creates a new action id with a generated identifier.

An action id with a new guid as the id. When proffering actions from an endpoint, prefer their ID. However, if you are manually creating your own actions, use this to easily generate a new ID.

Represents an agent mention in some source text, ex: @agent

Initializes a new instance of the class.

The span of the original mention location. The inputs that make up the mention.

Gets the of the agent mentioned.

Represents a single code snippet.

Initializes a new instance of the class.

The content of the code snippet.

Gets the content of the code snippet.

Represents a command mention in some source text, ex: /command

Initializes a new instance of the class.

The location of the mention in the original prompt text. The input parts of the mention.

Gets the name of the command.

Represents users action on a Copilot confirmation and indicates the current state.

Initializes a new instance of the class.

Confirmation identifier being responded to. The title or primary question of the confirmation.

Gets the title or primary question of the confirmation.

Gets the extended description of the confirmation to show to the user, if any. Supports Markdown.

Use this place to supply details to the user about what exact actions will be taken.

Gets the state of the confirmation request.

Gets the optional additional data attached to a confirmation.

Status of a Copilot confirmation request.

Base class for content actions like function calls or confirmations.

Initializes a new instance of the class.

The unique id for this action.

Gets the id for this action.

Represents a piece of content that is specifically code.

Initializes a new instance of the class.

The id of this content part. The code content.

Initializes a new instance of the class.

The code content.

Gets the language this code is in, if one has been provided. This is the formal name of the language, i.e. C#, Java, and others listed in .

If this is null, this was not available at the time of creation.

Gets the markdown identifier for the language this code is in, if one has been provided. Use this for Markdown code block rendering. This may be different from what is set as , depending on what Markdown uses.

If this is null, this was not available at the time of creation.

Gets a value indicating whether this code part is considered code to be applied.

Gets a file name or path if one is associated with this code content.

Gets the to apply to this code content.

Gets user provided commands that are associated with this code content.

Represents a command/action that can be attached to Copilot responses to be invoked by the end user. ..

Initializes a new instance of the class.

The localized display name of the command which will be shown on UI. The command invocation which describe the command.

Gets the localized display name of this command.

Gets the localized display description of this command.

Gets the command invocation.

A content part representing a confirmation for user to approve/reject for an action to be taken by the responder.

Initializes a new instance of the class.

The id of this content part. The confirmation.

Initializes a new instance of the class.

The confirmation.

Gets the confirmation instance for user to act on.

Represents a piece of content that contains a list of and an associated name for the list.

Initializes a new instance of the class.

The id of this content part. The name of the file list. The files in the list.

Initializes a new instance of the class.

The name of the file list. The files in the list.

Gets the name of the file list.

Gets the files in the list.

Represents a set of follow up prompts.

Initializes a new instance of the class.

The follow ups.

Initializes a new instance of the class.

The id of this content part. The follow ups.

Initializes a new instance of the class.

The content part id.

Gets the function invocation.

Gets the confirmation for this function if there was one.

Gets the confirmation requirement for this function call.

Gets the result of the function call.

Gets a value indicating whether this is the beginning of a function invocation.

Represents an image.

Initializes a new instance of the class.

The image data. The mime type of the image. The image width. The image height.

Initializes a new instance of the class.

The id for this content part. The image data. The mime type of the image. The image width. The image height.

Initializes a new instance of the class.

The id for this content part. The visibility of this content part. The image data. The mime type of the image. The image width. The image height.

Gets the data of the image.

Gets the mime type of the image.

Gets the image width.

Gets the image height.

Gets the base64 URI of the image.

This property is a computed property based off of the and properties. Once computed the value will be cached.

Represents an inline request to the Chat UI.

Gets the options of the inline request.

Gets a value indicating whether we will wait for the user action before continuing the conversation or not.

Gets or sets a value indicating whether the user completed (submit, cancel, approve, reject) the inline request.

Gets or sets a value indicating whether the user canceled or rejected the inline request.

Represents a section of Markdown formatted text content.

Initializes a new instance of the class.

The markdown content of this part.

Initializes a new instance of the class.

The id of this content part. The markdown content of this part.

Gets parsed mentions present in .

Content part that includes instructions coming from MCP servers.

Initializes a new instance of the class.

The instructions content. The name of the server the instructions come from.

Initializes a new instance of the class.

The id of this content part. The instruction content. The name of the server the instructions come from.

Gets the id of the content.

Gets the visibility of the content part for controlling what gets shown to the model or the user.

Gets the annotations associated with this part.

A message-local identifier for individual content parts.

Represents a progress item that will persist in its final form.

A progress part differs from setting a part status where only indicates whether the parts content is expected to update, a defines a content representation of an actual task. The task that occurred is the content, so the final state will be preserved and presented to the user.

Initializes a new instance of the class.

The id for this content part.

Gets the current state message, if any.

Gets an extended description of the current state, such as more details for what actions were taken, in limited Markdown.

Gets the status of this progress item.

Content part that contains reasoning information including any encrypted content that model sent.

Gets the model specified message identifier.

Gets the encrypted content sent by the model, if any.

Gets the reasoning content sent by the model, if any.

Gets the number of reasoning output tokens reported by the model for the originating message.

Gets the source API that generated this reasoning content.

Represents a summarization text piece of content.

Initializes a new instance of the class.

The text content of the part.

Initializes a new instance of the class.

The id for this content part. The text content of the part.

Gets the text content of this content part.

Represents a system diagnostic content part for debugging and error reporting.

System diagnostics are used to capture errors, warnings, and informational messages from deep in the Copilot stack and surface them to the UI for debugging purposes. By default, these parts have set to default (neither Model nor User), meaning they are not shown in normal UI rendering or sent to the model. They are intended for internal debugging and error reporting.

Initializes a new instance of the class.

The id for this content part.

Gets the severity level of the diagnostic.

Gets the source component that generated the diagnostic.

Examples: "LLMClient", "ResponderBroker", "Summarizer", "CopilotAPIClient".

Gets the diagnostic code for categorization.

Examples: "http_error", "token_budget_exceeded", "content_filter_triggered".

Gets the human-readable diagnostic message.

Gets additional structured data associated with the diagnostic.

Examples: HTTP status codes, token counts, exception types.

Gets the stack trace associated with the diagnostic, if available.

Represents a plain text piece of content.

Initializes a new instance of the class.

The text content of the part.

Initializes a new instance of the class.

The id for this content part. The text content of the part.

Gets the text content of this content part.

This content is visible to models and will be included in any submitted prompts.

This content is visible to the user and can be included in any UI representation of it.

This content is visible to both the user through the UI and to the model.

Represents a context mention in some source text, ex: #file:`File.cs`

Initializes a new instance of the class.

Gets a value indicating whether the context mention should be excluded or not

Indicates the severity level of a system diagnostic.

Informational diagnostic

Warning diagnostic

Error diagnostic, use to display diagnostic error messages

Represents a collection a code snippets associated with a file.

Gets the file name or path associated with this code content.

Gets the code snippets associated with this file.

Gets a value indicating whether the user has accepted at least one of the snippets.

Represents the arguments for a Copilot function.

Maps current arguments into a dictionary that abide by the given .

The input descriptors to map the arguments to. The mapped dictionary

Json version of arguments to be shown to a model.

The json version of the arguments.

Creates a new instance of from the given json.

The json to create the arguments from. The created arguments.

Creates a new instance of from the given object.

The object to create the arguments from. The created arguments.

Initializes a new instance of the class.

Json of the arguments

Represents a function invocation action.

Initializes a new instance of the class.

The id of this function invocation. The name of the function to invoke. The arguments to the function.

Initializes a new instance of the class.

The id of this function invocation. The name of the function to invoke.

Initializes a new instance of the class.

The name of the function to invoke. The arguments to the function.

Gets the name of this function.

Gets the arguments for this function.

Base class for mentions in request and response content.

This is the base class for the various types of mentionables: 1. Agents 2. Context 3. Commands 4. Variables Mentionables share a few characteristics: 1. They all have a leading syntax character. 2. They accept inputs. If the parsed mention type has a name, it will be in the first input. Mentions have two phases: Parsed and Queried. Parsing does not determine the presence of the mentioned item, only its syntax. Querying is implemented internally and takes a variety of paths to display or determine presence before invocation.

Initializes a new instance of the class.

Gets the span for the full mention content in the original text.

Gets the inputs that make up the mention.

Gets the leading symbol for this mention.

The leading symbol for context mentions: #

The leading symbol for scope mentions: #

The leading symbol for agent mentions: @

The leading symbol for command mentions: /

The leading symbol for external variables: !

The symbol to separate inputs: :

The quotation symbol for inputs: '

The input character escape symbol: \

These symbols are treated as punctuation so they will be escaped and end the parsing of a reference.

Creates the raw string syntax for a mention.

The type of mention. The input parts to the mention. The raw string format of the mention.

Takes an individual mention input and escapes it.

The original input. An escaped version of . is .

Takes an escaped individual mention input and returns the unescaped version.

The escaped input to unescape. An unescaped version of .

The type of mention.

The mention type is not yet known.

The mention refers to a piece of context or a scope.

The mention refers to a top level or agent command.

The mention refers to a specific agent.

The mention refers to a variable that a responder will handle.

Represents the progress of a work item (such as a function call).

Initializes a new instance of the class.

Gets the current state of progress to show to the user, if any.

Gets an extended description of the current state to show to the user (as limited Markdown), such as more details for what actions were taken.

Gets the progress status.

States for progress.

Work item is in unknown or progress state.

Work item is complete.

Work item exited with an error.

Represents a resolved agent mention.

Initializes a new instance of the class.

The that provided this queryable mention. The full expanded parseable string without trigger characters. The display name.

Represents a resolved command mention.

Initializes a new instance of the class.

The that provided this queryable mention. The name of this command.

Initializes a new instance of the class.

The that provided this queryable mention. The full expanded parseable string without trigger characters. The display name.

Represents a queried variable mention.

Initializes a new instance of the class.

The that provided this queryable mention. The full expanded parseable string without trigger characters. The display name.

Types of retry requests.

A standard retry request without modifications.

Indicates that the retry was triggered after modifying the request references.

Gets the span content from the original source text.

The source text to extract the span content from. The span content from the original source text.

Determines if a given span is within this span.

The span to check. if the given span is within this one.

Represents a variable mention in some source text, ex: !variable

Variables are agent-surfaced mentionables with no defined purpose. The agent can handle them in any way it likes.

Initializes a new instance of the class.

The span of the original mention location. The content of the mention.

Gets the contents of the variable.

Represents a piece of content used in a request/response interaction.

Represents a piece of content that contains text.

Gets text representation of the content.

Specifies the source API that generated the reasoning content.

Reasoning content from the Responses API.

Reasoning content from the Completions API.

Provides extension methods for augmenting instances with traits.

Asynchronously augments a single with a list of .

The to add traits into. The service broker to use for obtaining the . The cancellation token. A task that represents the asynchronous operation. The task result contains the augmented .

Asynchronously augments a single with a list of .

The list of to add traits into. The service broker to use for obtaining the . The cancellation token. A task that represents the asynchronous operation. The task result contains the augmented .

Represents an individual piece of context to include with a request or pass between functions.

Initializes a new instance of the class.

The moniker for the service. Context descriptor to retrieve member name and copilot type name. The associated request for this context. The actual serializable object representing the context.

Initializes a new instance of the class.

The moniker for the service. The specific member on the that this originates from. The associated request for this context. Copilot context value type name which could be a custom context type name or one of . The actual serializable object representing the context.

Initializes a new instance of the class.

Gets the identifier for this context.

Gets the name of the provider that generates and manages this context.

Gets the name of the source component that proffers this context.

Used mostly for validation and telemetry purposes.

Gets the relevance of this context

Primarily used to prioritize amongst multiple contexts.

Gets the name of the provider member this context is from.

Gets a value indicating whether the context is potentially reducible.

This can be speculative; if it is too expensive to determine this at context construction but an attempt can be made, set to . Context reduction is being deprecated in favor of summarization. This property will be removed in a future release.

Gets the request id for this context.

Gets an identifying reference present in a request that relates to this piece of context.

The value of a piece of context can vary dependent on its inputs.

Gets the list of traits associated with this context.

The number of traits discovered for a context can vary dependent on the trait providers available.

Gets a value indicating whether the context is ephemeral.

Ephemeral context is only included on the request it is originally generated for and is not re-attached to future chat requests as part of history. The context object will still be stored in history as a record of its existence in prior chat requests for rendering in the UI.

Describes an individual member of context.

Initializes a new instance of the class.

The name of the context member. A simple description of the member. The Copilot type name.

Initializes a new instance of the class.

The name of the context member. A model-visible description of the member. The Copilot type name. Potential inputs to constrain the values from this member; required inputs must appear before non-required inputs.

Initializes a new instance of the class.

The name of the context member. A simple description of the member. The Copilot type name.

Initializes a new instance of the class.

The name of the context member. A simple description of the member. The Copilot type name. Potential inputs to constrain the values from this member.

Gets the name of this piece of context.

Gets the description of what this piece of context provides.

Gets the data type that this context provides.

Gets the possible inputs that can be passed to this particular context member to augment the resulting value further.

For a document context, range might be an input to pre-constrain a document to only a selected region.

Represents an evaluation of context to its reduced and promptified form.

Initializes a new instance of the class.

The final list of utilized context. The full context prompt representing . The total token count of .

Initializes a new instance of the class.

List of prompt message and its associated context member type. The final list of utilized context. The full context prompt representing . The total token count of .

Gets the final reduced list of context objects that are represented in .

Gets list of context and their promptified content.

Represents an identifier for a single instance of context maintained through reduction.

Defines the relevance of a piece of context and how it was added to the request.

Unknown relevance.

The context was referred to the user by mention or attachment.

The context was inferred by heuristic or by function.

The context was included due to its relationship to the prompt or to other pieces of context.

The context was included as an indication of IDE or other state.

Defines the context for a diagnostic.

Initializes a new instance of the class.

The diagnostic file path. The error code of the diagnostic. The description of the diagnostic. The line where the diagnostic occurs.

Gets the file path for the diagnostic.

Gets the files paths for the diagnostic.

Gets the error code of the diagnostic.

Gets the description of the diagnostic.

Gets the line number of the diagnostic.

Gets the code snippet associated with the diagnostic.

Represents a query for potential matching mentionables.

Initializes a new instance of the class.

The type of query. The inputs into the query.

Gets the type of mentions being queried for.

Gets the ordered inputs from a query split from the separators.

Note that the first argument can either be the member being referenced (#file:) or the name of the search query (#File.cs). For the user input #file:'File.cs':15, the inputs will be: [0] = file [1] = File.cs [2] = 15

Gets the copilot session identifier if the completion belongs to a specific session.

Gets a value indicating whether the query needs to skip exclusion checks or not.

Temporary flag to enable skipping the exclusions,for the sake of performance, to be enabled only for UI purposes

Initializes a new instance of the class.

The model that the prompt will be sent to. The token budget for the prompt.

Gets the model used for this prompt.

Gets the token budget for this prompt.

Gets a value indicating whether to include ephemeral context in the prompt.

The default value is true, but when rendering a prompt for a prior (historical) request, this should be set to false.

Represents a piece of context that can requested later.

In order to get the actual object this queried item is to represent, the and are passed to .

Initializes a new instance of the class.

The context provider moniker providing this context. The member providing this context. The full expanded parseable string without trigger characters. The display name.

Initializes a new instance of the class.

This constructor handles building the full name of the context mention from the and . Inputs are processed in order as defined in the context descriptor until all are processed or the first non-required input is not found. The s in the inputs are stringified via . The context provider moniker providing this context. The descriptor of the member providing this context. The inputs used to generate this mention. The display name. Thrown when a required input is missing for the context descriptor. Thrown when an input value for the context descriptor is null.

Gets the member used to generate this context.

Gets a value indicating whether the referenced mention is excluded from the context.

Represents a mention that can be displayed or queried later.

Initializes a new instance of the class.

The provider moniker providing this mention. The full expanded parseable string without trigger characters. The display name.

Gets the type of mention.

Gets the name to display for this query item.

1. File.cs:15-20. 2. Exception The display name for a mentioned item should either represent the name/member of the mentionable being referenced (ex. exception) or the identifier for a particular set of inputs to a member (ex. File.cs). As such, the display name should not be localized as it is either the same name as a user recognizes, or it will match what is inserted into the prompt.

Gets the full string to be inserted into the prompt, minus trigger characters.

The string here must be fully parseable to the exact context.

Gets a small additional bit of description to display to the user.

Current document.

Gets optional more detailed information to display to the user in a tooltip when selected in the completions list, otherwise the display name plus the description is shown.

This value is displayed to the user, so ensure it is appropriately localized.

Gets an icon to display next to the query items.

Gets a value indicating whether the referenced mention can be navigated to/opened.

Gets a priority for this item. Defaults to .

Gets the name of the provider that generated this mention.

Gets the inputs used to generate this mention.

No priority.

Some prioritization, for items that currently visible.

Files that are open, but not focused.

High prioritization, for items that are an area of focus for the user.

The file of focus or the current exception.

Selection prioritization, for items that are selected by the user.

The section of a document highlighted by the user.

Represents a query into a scope.

Initializes a new instance of the class.

Options for guiding content or context reduction.

Context reduction is being deprecated in favor of summarization. This type will be removed in a future release.

Gets the number of tokens available.

Gets the model used for the request.

Since counting tokens to reduce the context depends on the model that the request was made with, this should be used to ensure the correct amount of tokens are counted.

Gets the copilot request.

Represents the base rendering output for a render call.

Represents the text rendering version of a Copilot rendering.

Gets the text rendering content.

Provider that provides relevant context for Copilot completions. Must be exported with the ContentType attribute for the content type it supports.

Gets the unique identifier for this context provider.

Gets a value indicating whether this context provider is enabled.

Retrieves a collection of context items asynchronously for a given document.

Provides the necessary information about the document for which context items are being retrieved. Allows the operation to be canceled if needed. An asynchronous enumerable collection of context items related to the specified document.

Represents a collection of relevant information needed to handle a Copilot completion context request.

The path of the document in which the completion was triggered. The cursor position within the document where the completion was triggered. Identifies the programming language of the document. Denotes the version number of the document. The snapshot for the document when the completion was triggered. Provider should use the content from snapshot if one is available. featureFlags that are current active. The value could be of type `string`, `int`, `bool` or `string[]`.

Represents a collection of relevant information needed to handle a Copilot completion context request.

The path of the document in which the completion was triggered.

The cursor position within the document where the completion was triggered.

Identifies the programming language of the document.

Denotes the version number of the document.

The snapshot for the document when the completion was triggered. Provider should use the content from snapshot if one is available.

featureFlags that are current active. The value could be of type `string`, `int`, `bool` or `string[]`.

The base type for a Copilot completion context item.

This interface is reserved for implementation by its associated APIs. We reserve the right to change it in the future.

Represents a trait item, which is a key-value pair used for short string snippets.

Specifies the name of the trait. Specifies the value of the trait. Specifies the relative importance with respect to items of the same type. Accepted values are integers in the range [0, 100], where 100 is the highest importance. Items with non-conforming importance values will be filtered out. Default value is 0.

Represents a trait item, which is a key-value pair used for short string snippets.

Specifies the name of the trait.

Specifies the value of the trait.

Specifies the relative importance with respect to items of the same type. Accepted values are integers in the range [0, 100], where 100 is the highest importance. Items with non-conforming importance values will be filtered out. Default value is 0.

Represents a code snippet item extracted from a file.

Uri of the file from which the code snippet is extracted from. Contains the actual code snippet text. Additional URIs that contribute the same code snippet. Specifies the relative importance with respect to items of the same type. Accepted values are integers in the range [0, 100], where 100 is the highest importance. Items with non-conforming importance values will be filtered out. Default value is 0.

Represents a code snippet item extracted from a file.

Uri of the file from which the code snippet is extracted from.

Contains the actual code snippet text.

Additional URIs that contribute the same code snippet.

Provides the list of related files to include additional context to the prompt for Copilot inline completions.

This is a MEF component part, and should be exported with the following attributes: A Name and a ContentType.

Gets a value indicating whether the provider is enabled.

Retrieves the list of related files of the active file.

The active file. An optional position that can be used to refine the calculation of related files. A cancellation token. The list of related files.

Represent a position in a text document expressed as zero-based line and zero-based character offset.

Line position in a document. Character offset on a line in a document.

Represent a position in a text document expressed as zero-based line and zero-based character offset.

Line position in a document. Character offset on a line in a document.

Line position in a document.

Character offset on a line in a document.

Represents a base interface for handling Copilot types. This interface is designed to be implemented by optional interfaces that extend its functionality to handle specific Copilot type scenarios.

Interface to extend the functionality of to provide parsing capabilities.

Parses the specified input into a based on the provided .

The Copilot type name to parse the input into. The input string to be parsed. A token to monitor for cancellation requests. A task that represents the asynchronous operation. The task result contains the parsed .

Represents a generic to render a value into a format to be sent to the model.

Renders the given type value into the format to be sent to the model.

The type value to be rendered. Cancellation token Rendered result

Interface to extend the functionality of to render a value into a format to be sent to the model.

A rich document context object.

Initializes a new instance of the class.

Content of the document The relevant skeleton code.

Initializes a new instance of the class.

Content of the document The relevant skeleton code.

Gets the relevant skeleton code of references in the document.

Gets the original skeleton code before any reductions. Empty if there has been no reduction.

Describes the kind of a from the customer's codebase.

Copied from GitHub Blackbird's Symbol Extractor, which largely does this same sort of consolidation of diverse concepts into a single language agnostic vocabulary. https://github.com/github/symbol-extraction/blob/main/crates/lib/src/lib.rs#L57C1-L89C1 I have intentionally retained names and members (sometimes commented out) to match symbol-extraction crate to make for an easier import of additional Tree-Sitter queries.

Describes the fully qualified name and location of a symbol definition in a file.

The name of the file in the repo. Generally relative to the repo or workspace root. The best guess of the fully qualified name of the symbol, as determined by syntax. The name of the symbol, excluding any namespaces. The range of the documentation comment, if any. The range of the name of the symbol in the document. The range of the function, class, or namespace body, if any. The full range of the declaration. The syntax kind of the element. is a best guess based on syntax alone and and may not precisely match the official language service's name for that symbol. It is intended to be used only for search and comparison within the indexing component.

Describes the fully qualified name and location of a symbol definition in a file.

The name of the file in the repo. Generally relative to the repo or workspace root.

The best guess of the fully qualified name of the symbol, as determined by syntax.

The name of the symbol, excluding any namespaces.

The range of the documentation comment, if any.

The range of the name of the symbol in the document.

The range of the function, class, or namespace body, if any.

The full range of the declaration.

The syntax kind of the element.

Represents a range of text in a document.

The start of the text. The length of the text in characters.

Represents a range of text in a document.

The start of the text. The length of the text in characters.

The start of the text.

The length of the text in characters.

An empty text range.

Gets the end of the text, relative to the start.

Determines if the given position is within the range.

The index of the character. true if the position is within this range.

Determines if the given range is within the range.

The range to check. true if the given range is within this one.

Creates a new range from the given start and end character positions.

The start character position. The end character position.

Slices the given to just the text within this range.

The text the range is constructed over. The text within this range, as a new, copied .

Slices the given to just the text within this range and reformats each line to have number of spaces of indentation.

The text the range is constructed over. The indentation to format the code with. The text within this range, as a new, copied .

Provides a formatter to handle inherited types that can not make use of .

The base type for the inherited types.

Initializes a new instance of the class.

The sub-types to support. The order of these types is a breaking change. By default, will be used to lookup a formatter.

Initializes a new instance of the class.

The formatter to use for the . The sub-types to support. The order of these types is a breaking change.

Attempt to deserialize an older version of the type before it was changed to an inheritance formatter.

Generate Serialize and Deserialize delegates.

The type to serialize/deserialize. is not assignable from . While we can call and get a formatter typed to our given , it only returns . We still need a way to call it, so this generates a delegate for which you can pass in and it will cast it and then call the strongly typed Serialize/Deserialize for the actual value type. This is done with expressions and a compile to save the reflection costs we'd otherwise need to incur for every serialize/deserialize of the type, which then necessitates storing the compiled pair in .

Represents a GitHub reference used by the GitHub Copilot Extensions framework.

This class should only be used if calling a responder that is a GitHub Copilot Extension.

Descriptor for remote agent context references.

Gets the type of the reference.

Gets the identifier of the reference.

Gets or sets the data of the reference, this can be a Json object including a string.

Gets or sets the serialized data of the reference, this field is used by MessagePack serialization.

Gets the metadata for the reference.

Gets a value indicating whether the reference is an implicitly added one.

Gets the display name of the reference if available.

Gets the display url of the reference if available.

Gets the tooltip/display text of the reference if available.

Gets the skill that needs to be requested in order to process this reference.

Initializes a new instance of the class.

The name of the scope.

Gets the name of the scope to serve as the prompt item (ex. #workspace).

Gets any inputs this scope accepts.

Represents the context of a symbol.

Initializes a new instance of the class.

Fully qualified name of the symbol. Unqualified name of the symbol. Code snippet of the symbol. of the symbol. Locations where the code snippet is sourced from.

Gets the fully qualified name of the symbol, e.g., Namespace.ClassName.SymbolName.

Gets the unqualified name of the symbol, e.g., SymbolName.

Gets the code snippet of the symbol.

Gets the of the symbol.

Gets the locations where the code snippet is sourced from.

Gets the file paths where the code snippet is sourced from.

Represents the location where a code snippet is sourced from.

The file path where the snippet is located. The span of the snippet within the file.

Represents the location where a code snippet is sourced from.

The file path where the snippet is located. The span of the snippet within the file.

The file path where the snippet is located.

The span of the snippet within the file.

Represents the type of the symbol.

A characteristic or property specific to a domain within the workspace. Traits are obtained from trait providers, which in turn will obtain them from available trait sources within a workspace.

Gets the provider moniker associated with the trait.

Gets the key of the trait.

Gets the value of the trait.

This is a string representation of the trait that is expected to be rendered in the prompt. It may contain some prompt engineering in regards on how the extender chooses to render it but they are not intended to be too descriptive or long. Traits that contain a valid value in will be rendered as follows: ## KNOWN TRAITS - trait1.Prompt - trait2.Prompt - traitN.Prompt ...

Initializes a new instance of the class.

The provider moniker associated with the trait. The key of the trait. The value of the trait which is expected to be rendered in the prompt.

A document for context.

Initializes a new instance of the class.

The content of the document.

Gets the contents of the document.

Gets the original content of the document before any changes or reduction. Null if the context has not been modified/reduced.

Gets any selected regions within the document.

Gets the file path to the document.

Gets the language of the document.

Gets the files paths for the document context.

A document region.

Initializes a new instance of the class.

The beginning position of the region. The end position of the region.

Gets a caret position if focused.

Gets the starting position of the region.

Gets the ending position of the region.

Represents a contract for batch reducing contexts. "/>.

Implement this interface to provide custom logic for reducing a set of contexts in batch. This is typically used in scenarios where multiple contexts need to be processed together to produce a more concise or relevant set of contexts. This interface is an optional extension to a context reducer. Context reduction is being deprecated in favor of summarization. This interface will be removed in a future release.

Provides the ability to reduce a set of contexts in batch.

The set of to reduce. Options for how to perform the reduction. A cancellation token. A new set of containing reduced . In order to properly represent the actual contexts that were used to the user, new context objects must be generated instead of simply returning new prompts. The parameter provides options for guiding the reduction process: - : Represents the number of tokens available for the reduction process. This helps ensure that the reduced context fits within the allowed token limit. - : Represents the language model used for the request. Different models may have different tokenization strategies and limits, so knowing the model helps tailor the reduction process. - : Represents the original user request, providing additional context or guidance for how the reduction should be performed. This can include user preferences or specific parts of the context that should be prioritized.

Copilot context files paths information.

A grouping of individual pieces of context, providing snapshotting.

Gets the complete set of members that context can be retrieved from.

This list should be a static definition, all members regardless of current state are declared here. If your context provider is for the current exception, you might have members like: CurrentException.Type CurrentException.Message CurrentException.StackTrace Alternatively, you may choose to bundle them into a single 'CurrentException' member that provides a summary of the current exception with a string.

Store and associate the current context with the .

Store may be called more than once with the same , any duplicate calls should no-op.

Free any context associated with .

Gets the currently available members that context can be retrieved from.

A list of the members available from the state associated with .

Gets the context from the associated with given the .

The request id to indicate the specific version of context to retrieve. The name of the member from . The inputs to the member from . A cancellation token. The or if the context could not be obtained. Do not free context once it has been requested here, only do so from . In some cases it be not worth the cost of preserving context should it change in-between a request start and stop. In these cases it is valid to return a context which will abort any dependent execution.

Gets the information needed to display these contexts to the user.

A batch of contexts to be converted to display info. A cancellation token.

A reducer for copilot contexts.

Context reduction is being deprecated in favor of summarization. This interface will be removed in a future release.

Attempt to reduce the contents of the context.

The context object to reduce. Options for how to perform the reduction. A token counter. A cancellation token. A new containing a smaller . In order to properly represent the actual context that was used to the user, a new context object must be generated instead of simply returning a new prompt.

Represents a debugger context provider.

Gets the state of the program paused in the debugger as context.

The correlation id for the request to use the context with. Cancellation token. The current debugger context if a program is paused in the debugger or .

Represents a diagnostic context provider.

Gets the diagnostic context for the active document.

The correlation ID. The cancellation token. The diagnostic context for the active document.

Gets the diagnostic context for a specific file.

The correlation ID. The path of the file. The cancellation token. The diagnostic context for the specified file.

Gets the diagnostic context for a specific file within a range of lines.

The correlation ID. The path of the file. The starting line number. The ending line number. The cancellation token. The diagnostic context for the specified file within the range of lines.

Gets the diagnostic context related to build issues for all files in the workspace.

The correlation ID. If true warnings will be included in addition errors as well. UTC time that the build ended to help with synchronization, this is optional and DateTime.Min can be passed instead to retrieve currently cached items. The cancellation token. The diagnostic context for all files related to build issues.

Represents a document context provider.

Gets a document context given a .

The correlation id for the request. File path of document context to acquire Cancellation token. A context object containing a document with given if there is one, or .

Gets the active document as context.

The correlation id for the request. Cancellation token. A context object containing the active document if there is one, or .

Gets an explicit selection by the user as context if the last active document had a selection.

The correlation id for the request. Cancellation token. A context object containing the active document with selection or if nothing is selected.

Provides functionality for generating, displaying and promptifying interactions.

Promptify a from an original or reduced

Value to promptify. Cancellation token to monitor. A prompt snippet that represents the promptified version.

Confirms a pending copilot call.

Promptifies the context of the given request.

The request to promptify the context of. Options like model and token budget for promptifying. Cancellation token. An evaluation of the context giving its final prompt, tokens and reduced context.

Asynchronously augments a single with a list of .

The to augment. The cancellation token. A task that represents the asynchronous operation. The task result contains the augmented .

Asynchronously augments a collection of instances with a list of .

The collection of instances to augment. The cancellation token. A task that represents the asynchronous operation. The task result contains the collection of augmented instances. The order of the contexts in the resulting collection will be the same as the order in the input collection.

Represents a provider of a mentionable components that can be queried for those mentions.

Your context provider or agent implements this in order to serve up completion items that it has to offer.

Query the mentionable provider for potential pieces of context or scopes with input based on .

The query to match against. Cancellation token A list of possible context that partially or fully matches . Empty if there are none.

Navigates to the mention in the UI. Invoked when the user clicks a navigable mention.

A mention to navigate to. A cancellation token. if the context was navigated to successfully, otherwise.

A 'scope' is a service that can provide related context to a request.

Gets the metadata for this scope.

Cancellation token. The information describing the scope.

Gets related context for the provided .

The copilot interaction that this scope is being asked about. Any inputs to the scope. Cancellation for querying the context. Context related to the . One scope's determination of relatability is not comparable to another, so scopes should order their return context from most to least related. Scopes should use their own heuristics to determine cut off points for the relatability of the context to return rather than simply returning everything ordered.

Provides a mechanism to count the tokens of a given string.

Count tokens in a piece of text.

The text needs to be tokenized. Cancellation token. The exact token count.

The function trimmed a piece of text by suffix based on token count needs to be preserved.

Text to be trimmed by suffix based on token count. Max token count to be preserved. Cancellation token. Trimmed string.

The function trimmed a piece of text by prefix based on token count needs to be preserved.

Text to be trimmed by prefix based on token count. Max token count to be preserved. Cancellation token. Trimmed string.

Represents a provider for Copilot traits.

Traits are characteristics or properties specific to a domain within the workspace. They are obtained from trait providers, which in turn obtain them from available trait sources within a workspace. For more details on traits, refer to the class. To register a trait provider, use the method from the interface. If the trait provider needs to provide workspace traits, it should be registered against context type.

Retrieves the traits for a given .

The context to obtain traits for. The cancellation token. A task that represents the asynchronous operation. The task result contains the list of Copilot traits.

Initializes a new instance of the class.

The possible supported operations on a collection.

One or more items were added.

One or more items were removed.

The items in the list were reset and replaced with a new list of items.

One or more items were replaced.

will contain the items which have been replaced in their new forms.

The arguments representing a change in the Copilot collection.

Initializes a new instance of the class.

The action. The affected items in the collection.

Gets the action.

Gets the items in the collection that were affected by the action. If it is , then this will contain the new list.

Options for making a confirmation request with functions or agents.

Initializes a new instance of the class.

Title of the confirmation. Optional description.

Gets the title of the confirmation request displayed to the user.

Gets the options description of the confirmation request displayed to the user.

Shared context member and input key names for Copilot context providers and consumers.

The key for file path input in context providers and completion items.

The member name for error context.

Initializes a new instance of the class. Represents a Context prompt with promptified content.

The context/ Promptified context. Promptified context, excluding headers.

Gets the context object.

Gets the context content promptification, excluding headers.

Serialization options for JSON schema generation.

The version that should be used when registering context (those that will use ).

This version number will increment whenever a breaking change is made so that it's clear to both sides which behavior is expected from the remote side.

The version that should be used when registering context (those that will use ).

This version number will increment whenever a breaking change is made so that it's clear to both sides which behavior is expected from the remote side.

The version that should be used when registering a responder (those that will use ).

This version number will increment whenever a breaking change is made so that it's clear to both sides which behavior is expected from the remote side.

The version that should be used when registering a trait provider (those that will use ).

This version number will increment whenever a breaking change is made so that it's clear to both sides which behavior is expected from the remote side.

The version that should be used when registering a type handler (those that will use ).

This version number will increment whenever a breaking change is made so that it's clear to both sides which behavior is expected from the remote side.

Gets the for the Visual Studio copilot service. Use the interface for the client proxy for this service.

Creates a descriptor for use for interacting with extensions.

The name on the moniker that will activate the context provider. The descriptor for use when requesting activation of the context provider or when proffering the context provider to a brokered service container.

Creates a descriptor for use for interacting with extensions.

The name on the moniker that will activate the trait provider. The descriptor for use when requesting activation of the trait provider or when proffering the trait provider to a brokered service container.

Creates a descriptor for use for interacting with extensions.

The moniker that will activate the context provider. The descriptor for use when requesting activation of the context provider or when proffering the context provider to a brokered service container.

Creates a descriptor for use for interacting with runtime registered services.

The moniker that will active the type provider. The descriptor to utilize.

Creates a descriptor for use for interacting with runtime registered services.

The name on the moniker that will activate the type handler. The descriptor for use when requesting activation of the type handler or when proffering the type handler to a brokered service container.

Creates a descriptor for use for interacting with runtime registered services.

The moniker that will activate the type handler. The descriptor for use when requesting activation of the type handler or when proffering the type handler to a brokered service container.

Gets the descriptor for the default agent for the chat window experience.

Options for making a direct request with.

Gets a value indicating whether to automatically generate active (streaming) responses.

Gets the active interaction listener override to use for the active responses.

This overrides the default active interaction listener for offering active responses for this request. must also be or it will be ignored as the broker will not offer active responses. This will not be called for requests.

Gets the history items to include with the request (not including the request); automatically includes all history.

Initializes a new instance of the class.

Used to construct a new instance by copying the base .

Service for managing the exclusion of content in Copilot Chat.

Checks if a given content is excluded.

The source of the content, e.g., file path, URL, text document identifier. Content to check for exclusion. Cancellation token. if the should be excluded, otherwise.

Checks if a specific file is excluded.

The path of the file to check for exclusion. Cancellation token. if the should be excluded, otherwise.

Checks if the provided workspace has any exclusion rules.

The path of the workspace to check for exclusion rules. Cancellation token. if the has any exclusion rules, otherwise. This function returns the same result as IsAnyExclusionRulesAsync

Checks if the active workspace has any exclusion rules.

Cancellation token. if the active workspace has any exclusion rules, otherwise.

Gets the excluded files in a collection of files.

A collection of files paths to check for exclusion. Cancellation token A collection of file path for the excluded files.

Checks if the context should be excluded by Copilot exclusion rules.

Copilot context to check for exclusion. Cancellation Token. if the should be excluded, otherwise.

Gets the excluded contexts for a collection of contexts.

A collection of copilot contexts to check for exclusion. Cancellation token A collection of for the contexts that should be excluded.

Screens the request's and returns a new version of the request with exclusions removed.

Copilot request. Cancellation Token A new version of after filtering out the exclusions. This method screen the request for exclusions and removes the excluded context altogether from the request therefore is best called just before turning things into a prompt.

Describes a piece of input to the model, such as arguments to a function.

Initializes a new instance of the class.

The model-visible name of the input. The description of the input. The type of the input.

Initializes a new instance of the class.

The model-visible name of the input. The description of the input. The type name of the input.

Initializes a new instance of the class for an input that fills a particular schema but may not have a known .NET type that conforms to it.

The model-visible name of the input. The already computed type name. Use from if you have an arbitrary JSON schema for which you haven't a .NET type. The JSON schema of the input.

Gets the model-visible name of the input.

Gets the (usually assembly qualified) full type name of the data type stored.

Gets a value indicating whether the input is required (defaults to true).

Gets the schema for the input.

Describes a piece of input to the model, such as arguments to a function.

Initializes a new instance of the class.

The model-visible name of the input. The model-visible description of the input. The Copilot type name of the input.

Gets the model-visible name of the input.

Gets the model-visible description of the input.

Gets the Copilot type name of the input.

Gets a value indicating whether gets whether or not the input is required input (defaults to true).

The intention of the associated request is to explain a specific subject specified in the content.

The intention of the associated request is to answer a general question or a specific question about provided context.

The intention of the associated request is to fix an issue with the provided code or context.

The intention of the associated request is to generate new code based on the content and any provided context.

The intention of the associated request is to modify the provided code or context.

THe intention of the associated request is to modify the provided code or context to improve its performance.

The intention of the associated request is to generate tests for the provided code or context.

The intention of the associated request is to get details on how specific task could be achieved in VS and supply the user with information about the VS Command to achieve that.

The intention of the associated request is to generate or modify code.

The intention of the associated request is to document code.

No intent. Pass the request directly through.

Warning: The None intent is not recommended for most scenarios. The request will be executed as-is from the caller with the bare minimum amount of additional processing. This means that the request will not have basic safety guardrails applied automatically to the system prompt. The caller is responsible for adding those. Additionally, core platform features like custom instructions will not be applied to the request. Consider using instead of None, since that provides a better foundation for many custom Copilot scenarios.

This is a generic intent that includes some core platform features without being opinionated on the exact user-facing scenario.

This intent will ensure the use of basic safety guardrails in the system prompt as well as opt in to core platform features such as custom instructions. This is a good intent to use for extenders who are not sure which intent to use, or for scenarios where you want to customize the system prompt with guidance for your specific scenario.

Do not use.

Gets the context provided or obtained for this interaction.

For requests, this context will include anything that was pre-supplied. For responses, this context will include anything that was supplied in the request, as well as any context obtained and included during the response pipeline.

Gets the ordered content parts of this interaction.

Gets the metadata associated with this interaction.

Gets the event message associated with this interaction.

Initializes a new instance of the class.

Who and how the event was created Content to be displayed to the user Whether the event message appears before or after the interaction

Gets the content of the event message in Markdown format.

Gets who created the event message, such as a user action or an agent response.

Gets where the event message should appear in relation to the interaction.

The event message appears before the interaction, above the message box.

The event message appears after the interaction, below the message box.

The action was created by the user, such as a button click or command execution.

The action was created by the system, such as an automatic response or notification.

Represents a JSON schema.

Gets or sets a string representation of a JSON schema.

Creates a from a , ensuring the schema is post-processed.

The JSON schema object to convert. Optional JSON serializer options for converting to string. A new with the post-processed schema. This method automatically calls JsonSchemaMapper.PostProcessSchema on the provided schema to ensure all object types have additionalProperties: false and all properties are marked as required.

Represents a model that inputs and outputs language.

The gpt-4o family name.

The gpt-4.1 family name.

The gpt-5 family name.

The gpt-5.2 family name.

The gpt-5-mini family name.

The gpt-5.1-codex-max family name.

The Claude Sonnet 4.5 family name.

The gpt-4o-mini family name.

The family name of the o1 model that is used for general availability versions of the o1 models.

The o3-mini family name.

Initializes a new instance of the class.

The raw identifier of the model. The name of the model, such as gpt-3.5-turbo-0613. The model family name, such as gpt-3.5-turbo. The number of tokens that can be provided for context (request, context, history).

Initializes a new instance of the class.

Gets the max number of tokens that can be provided for input (request, context, history).

Gets the name of the encoding used for tokenization. if the encoding name is unknown.

Initializes a new instance of the class.

The raw identifier of the model. The name of the model, such as gpt-3.5-turbo-0613. The display name of the model, such as "GPT 3.5 Turbo". The model family name, such as gpt-3.5-turbo. The capabilities of the model. The is an opaque string and should not be interpreted. For scenarios where you want to adjust prompts for different models or change any other behavior based on details about the model, you should use the and/or properties.

Gets the raw identifier of the model.

Gets the name of the model, such as gpt-3.5-turbo-0613.

Gets the display name of the model, such as "GPT 3.5 Turbo".

Gets the model family name, such as gpt-3.5-turbo.

Gets the capabilities of the model.

Gets the type of billing for this model, if any.

Gets or sets the endpoint configuration for the Copilot model.

Gets or sets the additional endpoints supported by this model.

Gets the availability state of this model.

Gets the terms to accept usage of this model.

Gets the SKUs that this model is restricted to. If this is empty, the model is available to all SKUs.

Gets the cost multiplier for this model.

Gets a value indicating whether whether the user is able to select this model.

Gets information messages regarding model availability.

Gets warning messages for the model to communicate model deprecation or changes to model pricing.

Gets the model tier which denotes its capabilities and usage (e.g., "powerful", "versatile", "lightweight").

Gets a value indicating for which purposes this model serves as the default.

In a set of models of a particular type (ex. ) from a responder, a model can be designated as the default model for a given purpose. If the model is the default chat model, it will be pre-selected in model picker UIs.

Gets a value indicating for which purposes this model serves as the fallback.

In a set of models of a particular type (ex. ) from a responder, a model can be designated as the fallback model for individual purposes. This model will be selected if the requested model has exceeded its quota.

Defines model billing types for Copilot.

No billing for this model.

Model is billed as a premium model.

Bring your own key models.

Auto model billing type.

Represents a collection of capabilities that a Copilot model supports as a bit field.

The model supports no additional capabilities (or the capabilities are unknown).

The model supports non-parallel tools/functions.

The model supports parallel tools/functions.

The model supports vision capabilities with image inputs.

Gets the default model for the broker.

The responder broker. The model's purpose. Gets the default model encoder for the broker. If no default model for the given was founnd in the list of models.

Gets the model for the model request from the broker.

The broker. The model request to determine the model with. The model from the broker matching the , or .

Determines whether this model is designated as the default chat model.

The model to check. True if this model is the default chat model; otherwise, false.

Creates a new model instance with the chat purpose added to the default purposes.

The model to modify. A new model instance marked as the default chat model.

Creates a new model instance with the chat purpose removed from the default purposes.

The model to modify. A new model instance no longer marked as the default chat model.

Represents the type of code/status for a model message.

Unknown or unrecognized message code for future compatibility.

Indicates the model is experiencing high usage and may encounter rate limits.

Indicates the model is degraded on GitHub's side.

Indicates the model is degraded on the provider's side.

Indicates the model is pending deprecation.

Represents a message containing code/status and its associated information or warning text. Ref: https://github.com/github/copilot/blob/337e789bfca1b35abd6853598d5a002b301024f8/docs/copilot-api-service/adr/0040-client-side-messaging.md.

The info or warning code defined by the GitHub Copilot API. The message text associated with the code. This message is provided by CAPI and is intended for display to users.

The info or warning code defined by the GitHub Copilot API.

The message text associated with the code. This message is provided by CAPI and is intended for display to users.

The purpose has not been defined or can not be determined.

Suited to chat experiences.

Represents a request for a that matches given conditions.

A request for a GPT-4o model.

A request for a GPT-4.1 model.

A request for a GPT-4o Mini model.

A request for the default chat model.

Initializes a new instance of the class.

The family of the model to request.

Initializes a new instance of the class.

The family of the model to request. The capabilities required for the model.

Initializes a new instance of the class with the specified purpose.

The purpose of the Copilot request.

Selects the first model that matches the conditions of this request.

The ordered list of models to select from. A model matching the selection criteria of this request, or if not found.

Selects the first model that matches the conditions of this request.

The ordered list of models to select from. A model matching the selection criteria of this request, or if not found.

Selects the first model that matches the conditions of this request.

The ordered list of models to select from. The current quotas to evaluate models against. A model matching the selection criteria of this request, or if not found.

Selects the first model that matches the conditions of this request.

The ordered list of models to select from. The current quotas to evaluate models against. Whether we are selecting a model for auto model. If true, we do not check quotas. A model matching the selection criteria of this request, or if not found.

The model is an unknown state. It may be available or may require configuration.

The model is available.

The model has been disabled. User configuration to enable may be possible.

The model is unavailable to the current user. This may be due to the user's subscription level and may require the user to upgrade before gaining access.

Initializes a new instance of the class.

Represents a prompt and the associated token count.

Initializes a new instance of the class.

The prompt. The total token count of .

Gets the prompt.

Gets the amount of tokens used by .

Represents a piece of a prompt meant for consumption by a model.

Initializes a new instance of the class.

The contents of this prompt snippet.

Gets the contents of the snippet.

Represents the quota information for services.

Initializes a new instance of the class.

The type of quota the data represents. The timestamp of this quota data in UTC. The time when the quota resets, if applicable. The maximum limit of the quota, if applicable. The current usage of the quota, if applicable. Indicates if further exceeding the quota limit is allowed.

Gets the type of quota the data represents.

Gets the timestamp of this quota data in UTC.

Gets when the quota will reset, if it resets, in UTC.

Gets the current usage of the quota if a limit exists.

Can be greater than .

Gets the maximum limit of the quota. if the quota has no limit.

Gets a value indicating whether the user is allowed to exceed the quota limit. It's possible for this value to change at any time for example if the user's organization policy changed or they maxed out on spending limits.

Gets the percentage of quota or rate limit used (0.0 to 100.0).

The quota is an unknown type.

The quota represents regular chat and can potentially represent others in the future.

The quota represents chat with premium features.

The quota represents completions.

Rate limit usage for the 7-day (weekly) window.

Rate limit usage for the 5-hour (session) window.

Options for a refinement session.

Gets the originating interaction ID that generated the change spawning the difference view.

Gets the default view mode for the difference view. If not defined, the view mode will start as SideBySide.

Gets the general intent of the request. enables automatic detection of intent.

Gets additional guidance for the handling of the request content.

Guidance will be provided to the model in additional to any guidance that may be supplied by way of a selected . You can consider this similar to setting or appending to the system prompt.

Gets the model selection request for this request.

Gets the list of responders this request is directed to if there is any. If list is empty, mentions and session participants will be used to determine the responders.

Gets the functions available to this request.

Local functions' lifetime is tied to the session instance, do not use the same instance across sessions.

Gets the name of the first command in the request.

The request to look for commands in. The of the first command in the request, or if there is none.

If request is a confirmation, gets the confirmation instance.

instance if request represents one, null otherwise.

Returns a new request with the added.

The request to add the functions to. The functions to add. A new instance of with added to .

Returns a new request with the local added.

The request to add the functions to. The functions to add. A new instance of with added to .

Returns a new request with the additional added.

The request to add the context to. The context to add. A new instance of with added to .

Returns a new request with the additional added.

The request to add the context to. The context to add. A new instance of with added to .

Returns a updated with the additional content .

The request to update with the extra content. The content part to add. The updated containing the new .

Represents the initiation type of a .

The user explicitly invoked this request.

The request was made without the user's explicit invocation.

Gets the tool mode to use for the .

Defaults to if there are any tools in .

Gets a value indicating what kind of retry request was executed .

Gets the CopilotCorrelationId of the retried . Returns if the request has not been retried.

Gets a value indicating whether the request can be summarized.

When set to , indicates that the content of this request can be summarized if needed.

Gets the maximum number of continuations (additional round trips to the model) to allow before forcing a request to finish.

If the value is or less than 0, the default maximum number of continuations is used. If the value is greater than or equal to 0, the lesser of this value and the default maximum number of continuations is used. Setting the value to 0 will not allow any continuations, so the first response will be returned immediately. If this value is set to 1, the request could have at most one continuation: * The first response where the model asks for a tool to be called * The client then calls the tool, collects the results, and calls back into the model (continues) with the tool results asking to finish the request with no further tool calls.

Gets the initiator of the , defaults to .

For requests that are the result of a user action explicitly to invoke Copilot, this value should be set to . Inside of a responder, this only applies to the main request made by the user. Sub-requests such as intent detection should set this to . For requests that are made without the user's explicit invocation of Copilot, such as rename suggestions, this value should be set to . Request continuations (such as tool calls or function confirmations) will ignore this setting and automatically utilize their subsequent requests. An attempt is made to detect sub-requests, but it is best to set the option explicitly. This does not represent a behavioral difference, it is only for billing purposes.

Initializes a new instance of the class.

Initializes a new instance of the class specifying the required tool.

The name of the required tool.

Represents a response to a copilot request.

A response instance is the final state of a response, it is no longer updating.

Gets the interaction ID.

Gets the author of this interaction.

Gets the final status of the response.

Gets the Copilot model used for the main request.

Gets any quota information that may be available from the responder's service.

A quota being present is not an indication that the quota type was utilized in the request.

Gets the example prompts that can be displayed below the response to ask follow up questions.

Is this essential for first release?

Gets a value indicating whether the interaction can be retried.

Generic error

Content filter error

Max tokens error

User does not have chat access

Not development related question

Content filter error due to public code match filter 'Snippy'.

Run out of quota for Copilot free.

Run out of quota for premium models.

Billing is not configured for the user's organization or account.

Initializes a new instance of the class.

Gets a value indicating whether to provide UI for this session.

When , this creates a new session for interacting with the underlying provider but does not create any UI for presenting it. When , this gets or creates a top-level session integrated to a provided UI. The nature of the top level UI is determined by the service, where it may be a single top-level chat session or it may create a new isolated chat session.

Gets the default chat model for the session.

The session. A cancellation token. Gets the default model encoder for the session. If no default chat language model was founnd in the list of models.

Gets the model for the model request from the session.

The session. The model request to determine the model with. A cancellation request. The model from the session matching the , or .

Initializes a new instance of the class.

The id of the session. The name for this session for UI display. Optional. The time the session was created. The end user who is chatting with Copilot. Additional metadata for the session.

Gets the name of the session.

Gets the session user.

Gets the time when the session is created.

Gets the id of the session.

Gets additional metadata for the session.

Gets the client id for the session.

Options used during the creation of a Copilot session.

The client Id to use for the session.

Options used during the creation of a Copilot session.

The client Id to use for the session.

Gets the default responder to use for the session. defaults to the built-in chat participant for selected experience.

Gets an optional name for the session. Used primarily for sessions that are attached to UI-visible threads.

Gets a value indicating whether the session should persist across restarts of the IDE. Defaults to false.

Gets the welcome screen options for the new session.

Gets the metadata to associate with the session.

Gets the history to pre-supply to the new session.

Gets the service that proffers this participant if any.

Generally, here would indicate the user and not null would point.

The session is currently available.

The session is unavailable because the user is currently signed out.

The session has been closed.

Represents a prompt and the associated token count.

Initializes a new instance of the class.

The prompt. The total token count of .

Represents an individual thread in the Copilot UI.

Initializes a new instance of the class.

The session backing the thread.

Gets the backing the thread.

Options for creating a new Copilot thread.

Threads have a HAS-A (not IS-A) relationship with sessions, so thread options have a HAS-A (not IS-A) relationship with session options.

Options for creating a new Copilot thread.

Threads have a HAS-A (not IS-A) relationship with sessions, so thread options have a HAS-A (not IS-A) relationship with session options.

Initializes a new instance of the class.

The identifier for the client of the thread.

Gets the identifier for the client of the thread.

Gets the mode indicating no tools will be used.

No tools in will be used when this mode is selected.

Gets the mode indicating tool usage is optional.

Because this mode makes tools optional, may contain zero or more tools.

Gets the mode indicating tool usage is required.

Because this mode requires usage of at least one tool, at least one tool must be provided in .

Gets the mode indicating usage of a specific tool is required.

Because this mode requires usage of a specific tool, must contain the required tool. The name of the required tool. An instance of requiring usage of a specific tool.

Identifies a type of some serializable value.

You may use to convert a to a . You may also use one of several well-known types from to get a .

Initializes a new instance of the class.

Gets the name of type (or the element type if an array.

Gets a value indicating whether or not the type is an array, then referring to the element type.

Gets the default type name for the given type.

, , derivative or a primitive type. The matching type name for built in types from , or .

A container for values used in or .

Initializes a new instance of the class.

Copilot context value type name which could be a custom context type name or one of . Serialized object value.

Initializes a new instance of the class.

Copilot context value type name which could be a custom context type name or one of . Value to be serialized, this value must be MessagePack serializable.

Gets the object.

Gets the Copilot type moniker of the value.

Gets the value of the context or null if value could not be deserialized or type is not available.

The object must be serializable via MessagePack and object value can not be null. For type checks, recommendation is to use instead as it will avoid unnecessary deserialization. Throw when value passed to init is null.

Tries to get the serialized value casting it to the provided type.

Type to cast value to. Reference to store the returned value. true if successful, false otherwise.

Options used during the creation of the welcome screen for a new Copilot session.

Dictionary key to use when storing these options in session metadata.

Gets the icon shown at the top of the welcome screen. Empty moniker means no icon is shown. Null means to use the default icon.

Gets the title override for the new session. Empty string means no title is shown. Null means to use the default title.

Gets a subtitle override for the new session. Empty string means no subtitle is shown. Null means to use the default subtitle.

Gets a value indicating whether discovery items are hidden.

Gets the ice breaker overrides for the new session. Null means to use the default ice breakers.

A context object containing the workspace traits context.

This context type's current only purpose is to serve as a placeholder to the workspace trait providers. This will only serve as a fallback for when no other context that provides traits is available.

Initializes a new instance of the class.

Get the encoder for this model if found.

The model to search for an associated encoder for. The cancellation token to use. The model associated encoder, or if one is not found.

Count the tokens for the .

The content to count. The total tokens takes.

Encodes the into tokens.

The content to encode into tokens. The list of tokens representing .

Get the text encoder for this language model, if supported.

The model to get the encoder for. The cancellation token to use. The encoder to use with the or if not supported.

Gets the text encoder for this language model if supported, or returns the default encoder.

The model to get the encoder for. The cancellation token to use. The encoder to use with the or a default encoder if one is not found for the provided model.

Microsoft.VisualStudio.Copilot experimental features diagnostic codes.

API features diagnostic code.

Internal features diagnostic code.

The confirmation requirements for a given action.

Confirmations are not required for this action.

Confirmations are required for this action.

Confirmations are always required for this action and may not be configured to be skipped by the user.

A model-visible description for the attributed element.

Initializes a new instance of the class.

The model-visible description for this element.

Initializes a new instance of the class.

The model-visible name for this element. The model-visible description from this element.

Gets the optional name override, visible to the model.

The name of the function or parameter attributed will be used by default.

Gets the model-visible description for this element.

Marks a method as a copilot function.

Initializes a new instance of the class.

The model-visible name for this element. The model-visible description from this element. The confirmation requirement.

Initializes a new instance of the class.

The model-visible description from this element. The confirmation requirement.

Gets the confirmation requirements for this function.

Gets or sets the name of a group that this function may be displayed within (e.g., "Planning").

Gets or sets a value indicating whether this function is only available during virtual function resolution.

When set to , this function will not be returned by and will not appear in the tool picker. It can only be accessed when a is resolved to this function.

Initializes a new instance of the class.

The model-facing name of the function. The model-facing description of the function. The confirmation requirements for this function.

Gets the model-facing name of the function.

Gets the model-facing description of the function.

Gets the name of a group that this function may be displayed within.

Gets the confirmation requirements for this function.

Gets the Copilot type of a return value of the function, if any.

Gets a value indicating whether this function is required by the responder that returns it. If true user will not be allowed to disable the function. If false the function will be enabled by default but user can disable it.

The default value is . This value is only applicable when the function is being returned by the active responder.

Gets a value indicating whether this function is hidden from function provisioning and only available during virtual function resolution.

Gets the arguments to the function.

Helper base class for function handlers to automate providing and invoking functions.

Methods on a subclass of this will automatically be listed so long as they are marked with . that this implements still expects to be an exported brokered service, so your subclass should be proffered as a brokered service matching . There is no need to otherwise register function providers.

Gets the brokered service descriptor for this function handler.

Functions generated from the subclass with be proffered as , pointing back to this brokered service.

Gets the functions from this provider, this signature includes resolution-only functions.

Correlation ID for the associated request. Cancellation token.

Resolves a virtual function descriptor to its concrete implementation based on the request context.

The virtual function descriptor to resolve. Correlation ID for the associated request. The functions that are marked as from the specified provider. The request properties used to conditionally select concrete tools. The property is empty to avoid leaking functions to implementation. Cancellation token. The resolved concrete function descriptor(s), or empty to use the fallback tool.

Contains the various extra properties that may be utilized by a function invocation.

Gets the function call ID.

Gets the session id that the request and response are part of.

Gets the interaction ID of the response that this function invocation is a part of.

Gets the original request that resulted in this function invocation.

Gets the progress broker for use by the function.

If is not invoked, a generic in-progress state before invocation and completed state after invocation will be used, if the call is presented to the user.

Gets the copilot function broker used for interacting with the client during function invocation.

Represents the identity of a Copilot function provider.

Initializes a new instance of the class with the specified display name.

The display name associated with the function provider identity.

Gets or initializes the display name of the function provider.

Gets or initializes the icon moniker associated with the function provider.

Represents the result of calling a function.

Creates a value function result from .

The value to return. A function value result containing .

Creates a value function result from .

The value to return. A function value result containing .

Creates a value function result from .

The value to return. A function value result containing .

Creates a value function result from .

The value to return. A function value result containing .

Creates a value function result from .

The value to return. A function value result containing .

Creates a value function result from .

The value to return. A function value result containing .

Creates a value function result from .

The value to return. A function value result containing .

Creates a value function result from .

The value to return. A function value result containing .

Creates a value function result from .

The value to return. A function value result containing .

Creates a value function result from .

The value to return. A function value result containing .

Creates a value function result from .

The value to return. A function value result containing .

Creates a value function result from .

The value to return. A function value result containing .

Creates a value function result from .

The value to return. A function value result containing .

Creates a value function result from .

The value to return. A function value result containing .

Creates a value function result from .

The values to return. A function value result containing .

Represents a data value returned from a function.

Initializes a new instance of the class.

The value returned from the function.

Gets the value returned from the function.

Represents a locally-provided function description.

Initializes a new instance of the class.

The model-facing name of the function. The model-facing description of the function. The confirmation requirements for this function.

Gets the local function invocation, if directly invokable.

Leave if you're providing a function you intend to handle by inspecting the incoming function part yourself.

Creates a local function descriptor from a delegate.

The function to create a description from. A local function descriptor. Throw if the given does not meet the constraints of a local function, such as using unsupported argument types or not including . Due to the limited RPC lifetime of a delegate created function, it's only recommended to use this for functions you're adding to your own requests (externally or inside an agent).

A descriptor for a Copilot function that is provided by an MCP server.

Initializes a new instance of the class.

The moniker by which the brokered service that can invoke this function may be requested. The name of the function as the server knows it. The path to the mcp.json file that describes the server that provides this function. The name of the function. This should be unique. Typically prefixed by a unique name of the server. The name of the function to display to the user. The description of the function. The confirmation requirement for the function.

Gets a value indicating whether the server configuration has placeholders for user-supplied inputs.

Represents a function originating from a .

Initializes a new instance of the class.

The service moniker to the that provides this function. The model-facing name of the function. The model-facing description of the function. The confirmation requirements for this function.

Gets the moniker for the providing this function.

Declares a virtual function that will be automatically included by . Apply this attribute to a class that extends to register virtual functions.

Initializes a new instance of the class.

Represents a virtual function descriptor that acts as a placeholder for model-specific implementations.

Virtual functions are display-only tools shown in the tool picker that get resolved to actual implementations based on model capabilities or other context. They are never sent directly to the LLM. Resolution is handled by the service identified by . The service's method is called to determine which concrete tools to use. If no custom resolution is provided, is used.

Initializes a new instance of the class.

The model-facing name of the virtual function. The model-facing description of the virtual function. The name of the tool to use when no resolver is available. The service moniker that handles resolution of this virtual function. Thrown if is null or whitespace. Thrown if is null.

Gets the name of the tool to use as fallback when virtual tool isn't resolved before sending the request.

Gets the service moniker that handles resolution of this virtual function.

The identified service's method is called to resolve this virtual function to concrete implementations based on model capabilities or other request context.

A marshaled Copilot function.

Generally not meant to be implemented directly as it mainly serves to underpin . Due to the RPC lifetime involved, it's not generally recommended to use for anything but directly supplying requests you are making.

Invokes the Copilot function.

The various contextual information that may be relevant to the function. Arguments as defined in the descriptors arguments and supplied by the invocation. Cancellation token. A value if the underling method has a return type.

Yields the function call to ask confirmation from the user for an operation.

Options for the confirmation request. Cancellation token to monitor. User's response to the confirmation. Thrown when a function asks for more then one confirmation in the same invocation.

During a function call, ask the user for input.

Options for the input request. Cancellation token to monitor User's response to the request.

A function handling endpoint.

Invoke the provided .

The details of the function invocation. The various contextual information that may be relevant to the function. A cancellation token. The return value of the function, if any. If is not invoked, a generic in-progress state before invocation and completed state after invocation will be used, if the call is presented to the user.

Standalone or optional agent component that can provide functions.

Gets the functions from this provider.

Caching the is recommended for additional performance benefits. Correlation ID for the associated request. Cancellation token. Functions from this provider.

Gets the initial list of functions applicable to all requests and subscribes to further changes.

Observer to listen for changes. Cancellation token to monitor. an that should be disposed to release the observer. Unlike , this method returns functions that are applicable to all requests.

Describes the status of an MCP server.

Gets the current status of the MCP server.

Gets the reason why the entity is disabled, if available.

Gets the path to an mcp.json file that disables this server.

Gets the path to an mcp.json file with higher precedence and an MCP server configuration with a colliding ID.

Gets a message indicating why the server cannot be started, if applicable.

Gets the number of inputs the server offers.

Gets the number of tools the server offers.

Gets the number of prompts the server offers.

Gets the number of resources the server offers.

Gets a value indicating whether the MCP feature is enabled.

Gets a value indicating whether the server can be started.

Gets a value indicating whether the server can be suspended.

Gets a value indicating the URI of the server, if applicable.

An enumeration of the possible statuses of an MCP server.

The server has not successfully started yet.

Indicates that the user must provide input before the server can start.

Indicates that authentication is required before the server can start.

A connection to the server is being established.

The server is running and ready to accept requests.

The connection to the server is temporarily suspended, but it can be resumed on demand by the LLM.

The server failed to start on the last attempt.

Specifies that the server is disabled.

A listener for active interactions to return a streaming interface.

Offers an active interaction to the listener who can optionally return a sink.

The author of the interaction. The correlation id of the interaction. The message id of the individual interaction. Cancellation token. Optionally returns a context containing the associated and an sink. This is called by a responder or broker to notify any listening party of an interaction that is to be created or updated. Typically the listener is the UI whereby it can then receive streaming updates from the responder. The to be returned in the context is the base contract. A request can return a , but its purpose is largely limited to updating the set of context that is being included. For example, if an agent adds context beyond what the user adds, it can call to update the list. For streaming responses, should instead be returned as it provides .

Service for managing Copilot component registrations.

Registers a copilot context provider.

The moniker that may be used to activate the context provider when needed. A cancellation token.

Registers a copilot scope.

The moniker that may be used to activate the scope when needed. A cancellation token.

Registers a instance at runtime.

Service moniker used to activate the type provider. Copilot type names supported by this provider. Cancellation token to monitor. This method should be used to register dynamic type providers that are not possible to declare ahead of the time.

Registers a for the specified Copilot types.

The service moniker used to activate the type handler. The list of Copilot type names that this handler supports. A cancellation token to monitor for cancellation requests. Type handlers will only get called for registered copilot type names.

Registers a .

The moniker that may be used to activate the trait provider when needed. The list of context types we want to support for this trait provider. Refer to for a list of available context values. A cancellation token. This method registers a with the specified provider moniker and context types. So the trait providers will only get called by the registered context types.

Registers an agent with the provided metadata.

Information required to register an agent. The agent service will not be queried until it is added to a conversation. Cancellation token to monitor.

Registers an agent with the provided metadata and options.

Information required to register an agent. The agent service will not be queried until it is added to a conversation. Options for agent interaction with Copilot chat infrastructure such as implicit context. Cancellation token to monitor.

Unregisters an agent that was assigned the provided moniker from the service.

Agent identity to unregister. Cancellation token to monitor.

Registers a function provider for the specified service moniker and function provider identity. Registered function providers can contribute to core chat infrastructure and tool picker UX.

Service moniker of the function provider. Identity of the function provider. Cancellation token to monitor for cancellation. Registration is optional, function providers used by private responders do not have to be registered. A task that represents the asynchronous operation.

Unregisters a previously registered function provider associated with the specified service moniker.

The unique identifier of the service whose function provider is to be unregistered. A token to monitor for cancellation requests. A task that represents the asynchronous operation.

Visual Studio service that can be queried via service broker.

Subscribes to availability status notifications.

An observer that will receive notifications based on availability. An initial notification will be sent immediately to inform of the current state. A cancellation token. A value that may be disposed of to cancel the subscription. When the service is not available, any existing becomes invalid, even if the service was available before and after an unavailability event.

Checks whether the conversation service is currently available.

A cancellation token. .

Gets or creates a copilot session.

Options to apply to the session. A cancellation token. Copilot session data. It must be disposed of by the caller. Thrown when the service is not available.

Retrieves a collection of quota information for the Copilot service.

Cancellation token. Quotas for the applicable s.

Gets or creates a copilot session without UI.

Options to apply to the session. A cancellation token. Copilot session data. It must be disposed of by the caller. Thrown when the service is not available.

Starts a new thread in the Copilot chat UI.

The options to start the thread with. Cancellation token.

Starts an inline refinement session with the proposed edits.

The initial edits to propose. Cancellation token. Starting a refinement session will cancel any existing refinement or inline session.

Starts an inline refinement session with the proposed edits.

The initial edits to propose. The ID of the interaction from which the code suggestion originated from. Cancellation token. Starting a refinement session will cancel any existing refinement or inline session.

Starts an inline refinement session with the proposed edits.

The initial edits to propose. Allows to customize refinement session behavior. Cancellation token. Starting a refinement session will cancel any existing refinement or inline session.

Attempt to show the UI associated with the provided .

The session to display the UI for. Cancellation token. For a session that has already been closed, or if a session that was not created with UI, this will no-op.

Opens the output window for the Copilot service.

Get all current sessions

Options to apply to the session. A cancellation token. List of sessions

Get all current threads.

The persistence path for the threads to return. A cancellation token. List of threads The caller is responsible for disposing the returned instances.

Get all current threads.

A cancellation token. List of threads The caller is responsible for disposing the returned instances.

Gets a session given a copilot session identifier, if there is no such session this method will return null.

Session identifier. A cancellation token. A instance if a session with the given id is available, null otherwise. This method returns a wrapper that is disposable only for purposes of disposing the proxy and will not dispose the actual session.

Starts an inline session at the start position of point.

The position to show inline session. The identifier for the client of the session. A cancellation token.

End a copilot session.

Session identifier. A cancellation token. No-op when deleting a non-existent session.

Get Copilot session info of this session (user, time created and id).

A cancellation token. the session info.

Subscribes to the interactions (queries and responses) in this session.

Observers will be sent the current list of interactions (if non-empty), and lists of new interactions that were appended or removed.

Gets a snapshot of the interactions up to this point.

A cancellation token. A snapshot of the interactions.

Ends the for all the users.

The cancellation token. Unlike Dispose(), which only cleans up the proxy for the caller, this method permanently ends the session for all proxy instances that point to the same underlying session. Call any methods after result in an being thrown.

Continues generating the given .

The response to continue. The request options, which should match those of the original . Cancellation token. The updated response. This is used to continue generating responses after a response has been updated with a manually supplied function result, or after a function confirmation has been received. doesn't come from this session or match a previously started response. or is .

Registers a listener that may hook up to live stream interactions.

The listener. A cancellation token. The async operation.

Gets the responders attached to the session.

A cancellation token. A list of responders attached to the session with the default in the first position.

Adds a responder provider to this session.

The service moniker for the service. A cancellation token. The async operation

Gets the models from the default responder of this session.

A cancellation token. A list of models from the default responder.

Gets the models from the specified responder provider.

The service moniker for the service. A cancellation token. A list of models from the responder.

Adds an additional Copilot model to the system asynchronously.

The Copilot model to be added. Cannot be null. A token that can be used to cancel the operation. If cancellation is requested, the operation will terminate early. A task that represents the asynchronous operation.

Removes the specified additional model from the system asynchronously.

The instance to be removed. This parameter cannot be null. A that can be used to cancel the operation. A task that represents the asynchronous operation.

Loads stored model providers from the configuration file.

List of .

Saves the given model providers to the configuration file. Retries up to 3 times before failing.

The providers to save.

Removes the interactions with the given message ids from the session.

The ids of the interactions to remove. A cancellation token.

Converts a context data object into a snippet of a prompt to be sent to a model.

Type and object to convert into a prompt snippet. Cancellation token. A model prompt representation of the .

A string which identifies a client for a session. This should be a dotted string similar to a namespace specifier, such as "Company.Product.Team.Feature[.Subfeature]".

Initializes a new instance of the class.

The string which identifies a client for a session.

Gets the client identifier.

A guid which links to the origin of a request and any responses related to it. Each initiated request has a unique CorrelationId.

The guid that correlates to the originating request.

A guid which links to the origin of a request and any responses related to it. Each initiated request has a unique CorrelationId.

The guid that correlates to the originating request.

Represents the identifiers for a specific Copilot interaction, whether active or finished.

The id for the conversation this interaction was a part of. The correlation id for this interaction. This id will match between requests and any responses to those requests. The id for the specific message. A is associated with a session (e.g. the visible chat window). A is a shared ID between a request and its response(s). A is given to each request and each response.

Represents the identifiers for a specific Copilot interaction, whether active or finished.

The id for the conversation this interaction was a part of.

The correlation id for this interaction. This id will match between requests and any responses to those requests.

The id for the specific message.

Initializes a new instance of the struct. for an existing conversation, where new correlation and message IDs will be generated.

Initializes a new instance of the struct for an existing conversation and correlation ID, where a new message ID will be generated.

Gets the id of the client for the interaction.

Identifies an individual message associated with a

A guid uniquely identifying a 'Conversation' that exists in the service. These are created by the service at the time a Conversation is initiated.

Initializes a new instance of the struct.

The guid that uniquely identifies the conversation. The identifier for the client of the conversation.

Gets the guid that uniquely identifies the conversation.

Gets the identifier for the client of the conversation.

Represents options for a actions-based inline request. This request option allows to display up to two commands to the user within Copilot Chat UI.

The commands can either be blocking or non-blocking. Blocking: The LLM will continue showing results from the user query. However, the user won't be able to submit a new query until they clicked one of the command. Non-Blocking: The LLM will continue showing results from the user query. User can completely ignore these commands and never click them.

Gets the first command (required). The key is the internal name and the value is the display name.

Gets the optional second action. The key is the internal name and the value is the display name.

Options for inline requests that include multiple fields.

Gets the dictionary of fields where the key is the internal name and the value is the corresponding UI element.

Gets a value indicating whether the current request is an approval request. This will impact how the submission buttons are presented. They can be presented as "Approve" and "Reject" instead of "Submit" and "Cancel".

Gets a value indicating whether the request has been pre-approved. When true, the form for the fields won't be displayed to the user, and the request will be submitted automatically with default values. If a field is required and does not have a valid default value, the request will fail.

Base class for options used to configure input elements in an inline request UI.

Gets an optional message to display to the user when prompting for input.

Result of an inline request to the user.

Gets the confirmation status of the inline request.

Gets the fields provided by the user in response to the inline request.

This is typically omitted if the is or . Values in the dictionary should be a JSON serialized String, Number or Boolean.

Represents a Boolean input element in an inline request UI.

Gets a value indicating whether the default value is true or false.

Represents an enumeration input element in an inline request UI.

Gets the list of enumeration values where the key is the internal name and the value is the display name.

Gets the default selected key for the enumeration.

Represents a label element in an inline request UI. The is ignored.

Represents a numeric input element in an inline request UI.

Gets the minimum allowed value.

Gets the maximum allowed value.

Gets a value indicating whether the number is expected to be an integer.

Gets the default value.

Represents a string input element in an inline request UI.

Gets the minimum length for the string.

Gets the maximum length for the string.

Gets a specific format for the string.

Gets a value indicating whether the element accepts multiple lines text input.

Gets a value indicating whether the element is read-only.

Gets the default value to display in the input element.

Defines the format of a string input in an inline request UI.

Base class for different types of input elements in an inline request UI. These models are on pair with MCP Elicitation API models, but aren't limited to. They can be extended as needed as long as we ensure the MCP parts are correctly mapped. See https://modelcontextprotocol.io/specification/draft/client/elicitation

Gets the title of the input element.

Gets the description of the input element.

Gets a value indicating whether the input element is required.

Represents a progress that can be waited upon for all progress events to finish.

The type of progress update value.

Returns a that completes when all currently pending reports have completed.

A cancellation token. A that completes when all currently pending reports have completed.

Marks the progress as complete and returns a that completes when all progress events have finished.

A that completes when all queued progress events have finished. Calls to after this is called will be ignored. Exceptions raised from will be captured in the returned .

Marks the progress as complete but does not wait for processing of remaining items..

A class that provides constants for common language names.

Parses the language name.

The language to parse. The language parsed, or null, if not supported.

Specific identifiers for rendering Markdown code blocks.

Resolves the language to an identifier to use in rendering Markdown code block.

The language. The identifier to use in the code block, or null if not supported.

Gets the status of a specific MCP server.

The full path to the mcp.json file. The name of the server, if filtering to a specific server is required. The observer that will receive status updates. A cancellation token. A value to dispose to cancel the subscription. Thrown when does not point to an mcp.json file that is being tracked. Thrown when the designated server cannot be found in the mcp.json file specified by .

Restarts the specified server.

The name of the server to restart. Cancellation token. A task that completes when the restart operation is done.

Suspends the specified server.

The name of the server to suspend. Cancellation token. A task that completes when the suspend operation is done.

Gets a value indicating whether the specified mcp.json file is being tracked by the MCP service.

The full path to the mcp.json file. A cancellation token. if the mcp.json file is relevant to the current workspace; otherwise . This returns for every file when the MCP feature is disabled.

Gets the number of inputs relevant to an mcp.json file, and optionally to a particular MCP server.

The full path to the mcp.json file. The name of the server, if filtering to a specific server is required. A cancellation token. The number of inputs the server configuration requires.

Enables the MCP feature, allowing MCP servers to be started.

Cancellation token. A task that completes when the enable operation is done.

Disables the MCP feature, preventing MCP servers from being started.

Cancellation token. A task that completes when the disable operation is done.

Brokered service for aggregating and discovering MCP configuration sources. Provides read-only aggregate views and access to individual sources.

Gets a list of all available configuration source descriptors, in precedence order.

Gets a dictionary of all servers across all configuration sources.

Gets the configuration for a specific source descriptor.

Subscribes to configuration change events for any source. The observer is notified when any underlying source changes.

The observer to notify on any configuration change. Cancellation token. A disposable for unsubscribing.

Subscribes to changes in the set of available sources (e.g., new sources discovered or removed). The observer is notified when the list of sources changes.

The observer to notify on source list changes. Cancellation token. A disposable for unsubscribing.

Gets a writable proxy for a specific configuration source.

The descriptor of the source to get Cancellation token. A proxy to the source that can be used for updating the configuration

Brokered service for accessing MCP registry policy configuration.

Gets the current MCP registry Configuration.

Subscribes to configuration change events to the current MCP registry. The observer is notified when the current MCP registry changes.

The observer to notify on any configuration change. Cancellation token. A disposable for unsubscribing.

Brokered service for accessing MCP Resources and MCP Resource Templates.

Gets a list of all available resource and resource template descriptors.

Gets the initial list of resource and resource template applicable to all requests and subscribes to further changes.

Observer to listen for changes. Cancellation token to monitor. an that should be disposed to release the observer.

Retrieves a specific MCP Resource by its resource ID.

Represents a writable configuration source for MCP.

Retrieves the current MCP configuration.

Applies the specified MCP configuration on the source. If a configuration already exists, it will be replaced with the provided configuration.

The new to apply A token to monitor for cancellation requests.

Gets inputs that are intrinsic to the environment or source of the configuration and not actually serialized in the mcp.json file.

Creates a dictionary of input IDs and the IDs of the servers that consume them.

Adds a new server to the configuration.

The unique identifier for the server. The server instance to add. A new instance with the added server. Thrown if a server with the specified ID already exists.

Removes a server from the configuration.

The unique identifier of the server to remove. A new instance with the server removed. Thrown if the server with the specified ID does not exist.

Updates an existing server in the configuration.

The unique identifier of the server to update. The updated server instance. A new instance with the updated server. Thrown if the server with the specified ID does not exist.

Adds a new input to the configuration.

The input to add. A new instance with the added input. Thrown if an input with the specified ID already exists.

Removes an input from the configuration.

The unique identifier of the input to remove. A new instance with the input removed. Thrown if the input with the specified ID does not exist.

Updates an existing input in the configuration.

The updated input instance. A new instance with the updated input. Thrown if the input with the specified ID does not exist.

Adds the inputs and servers from another configuration to this configuration. If there are ID conflicts, updates the configurations from the other to have unique IDs, including replacing any tokens in the server configuration with the new input IDs.

The other configuration to merge. A new instance with merged inputs and servers. Always thrown; not yet implemented.

Immutable, serializable descriptor for an MCP configuration source.

Serializable descriptor for a configuration retrieved from a specific MCP source.

Describes the source of the MCP configuration. The configuration that was retrieved from the source.

Serializable descriptor for a configuration retrieved from a specific MCP source.

Describes the source of the MCP configuration. The configuration that was retrieved from the source.

Describes the source of the MCP configuration.

The configuration that was retrieved from the source.

Gets the input's id is used to associate an input with a variable of the form ${input:id}.

Gets the type of user input prompt to use.

Gets the description is shown when the user is prompted for input.

Gets the default value is used if the user does not provide a value.

Gets a value indicating whether if true, the input is treated as a password and the value is hidden.

Gets the description is shown when the user is prompted for input.

Gets the default value is used if the user does not provide a value.

Gets the options are shown to the user when they are prompted for input.

Gets the command to execute for this input variable.

Gets optional arguments passed to the command.

May be either a or an array of strings.

Gets the label is shown to the user when they are prompted for input.

Gets the value is used when the user selects this option.

Represents an MCP Prompt and its descriptor.

A descriptor for an MCP prompt that is provided by an MCP server.

A descriptor for an argument of an MCP prompt.

Defines the access levels for the MCP registry.

Any server may run.

Only registry servers can run.

Represents the MCP registry configuration.

Gets the effective MCP registry policy for the signed-in user, if any.

Represents the policy for accessing an MCP registry.

Gets the base URL of the MCP registry.

Gets the access policy for the MCP registry.

Gets information about the owner of the MCP registry.

Represents an MCP Resource and its description.

Get name to be displayed in the attachment.

A descriptor for an MCP resource or resource template that is provided by an MCP server.

Represents metadata for a MCP server coming from vsweb+mcp:/install protocol.

Gets the name of the MCP server.

Gets the information about the MCP server to install.

Gets the version of the MCP server. This is optional and may be null.

Gets the list of input variables for the MCP server.

Gets path to a file containing environment variables for the server.

Specifies when an agent should be called to respond.

Call the agent only when it is the default responder or is mentioned directly.

Identifies an agent.

Initializes a new instance of the class.

The service. The short name used for mentions. The name to display for responses from this agent. The avatar used to to display for responses from this agent.

Gets the method as to when this agent should be called.

Gets the short name of the agent used to mention it. (ex. "foo" for @foo).

Gets a short user-visible description of the agent for display in completion.

Gets a longer user-visible description of the agent for display when users choose an agent starting a new thread.

Addition options defining how agents interact with chat.

Gets the default instance of options.

Gets a value indicating whether the agent supports implicit context.

Gets a value indicating whether the agent supports model selection.

A simple means to provide an existing builder for a call to .

It may be that a responder wants to manually assemble a response to the user, but still allow the broker to stream to the sink directly at a later stage.


             public async IAsyncEnumerable<CopilotResponse> RespondAsync(ICopilotResponderBroker broker, CopilotRequest request, CopilotRespondArgs respondArgs, [EnumeratorCancellation] CancellationToken cancellation)
             {
                 // Create a response to manually stream some content to the user.
                 ICopilotResponseBuilder builder = await broker.CreateResponseAsync(cancellation);
                 var getInfoProgress = await responseBuilder.CreateProgressContentAsync(cancellation);
                 getInfoProgress.Report("Getting information...");
                 // get extra info
                 getInfoProgress.Report("Got information", CopilotProgressStatus.Completed);
            
                 request = request with { /* add extra info */ };
            
                 var options = new CopilotDirectResponseOptions
                 {
                     // Use CopilotContinuationListener to provide the existing builder to the broker to stream to.
                     ActiveResponseListener = new CopilotContinuationListener(this.sessionId, request.CorrelationId, builder),
                 };
            
                 await foreach (var response in broker.SendDirectRequestAsync(finalRequest, options, cancellation))
                     yield return response;
             }

Initializes a new instance of the class.

The session id. The correlation id of the in-progress response. The in-progress response builder.

Provides arguments for call.

Gets the request options for the call.

Initializes a new instance of the class.

Gets the default chat model for the broker.

The responder broker. A cancellation token. Gets the default model for the broker. If no default chat model from the broker was found.

Gets the model for the model request from the broker.

The broker. The model request to determine the model with. A cancellation request. The model from the broker matching the , or .

Provides arguments for call.

Gets the request that is about to be sent to the Copilot service.

Gets the history of interactions in the conversation up to this point.

Initializes a new instance of the class.

The responder provider service that owns this responder. The full displayed name of the responder. The image moniker for the responders avatar.

Gets the responder provider service that owns this responder.

Gets a value indicating whether the device supports the MCP (Model-Context Protocol) feature.

Represents a Copilot session responder that supports interactive prompts during a request.

Respond to a given .

The broker for the responder to issue model requests through. Dispose before exiting. The request. The response that is being resumed. This may contain any user interactions such as confirmations. Arguments defining how the responder should behave. Cancellation token. 0 or more completed . is a proxy and must be disposed before exiting.

Interface for checking model availability in Copilot.

Gets the models available through Copilot.

Cancellation token. The models that Copilot supports.

Represents a Copilot session responder.

Respond to a given .

The broker for the responder to issue model requests through. Dispose before exiting. The request. Arguments defining how the responder should behave. Cancellation token. 0 or more completed . is a proxy and must be disposed before exiting.

A broker offering base services to a responder.

Brokers are created on a per-responder basis; the broker a responder receives is associated with their identity.

Sends a direct request to the model.

The request to send. Options for making the request. Cancellation token. 1 or more completed .

Creates a response builder for manually streaming and generating a response.

Cancellation token. A response builder. Calling this will cause the parent to proffer a . For sessions that are UI based, this means that pushing content to the returned builder will stream it directly to the UI. It is expected that is called and that a matching ( and ) is returned from .

Gets the history that applies to you for the session you're participating in.

Cancellation token. A list of your interactions in the session, not including the current request. For directly mentioned agents, this history is straightforward, only requests that mentioned the agent and agent responses will be included. For unmentioned responders, these will include any request that was responded to by that responder.

Continues a request that was paused previously for an user interaction such a confirmation prompt.

The request to send. The response to continue. Options for making the continuation request. Cancellation token. 1 or more completed .

Asynchronously resolves a collection of function descriptors that match the specified request criteria.

This method may throw an exception if the request is invalid or if the operation is canceled. The request object that defines the criteria for resolving functions. Cannot be null. A token that can be used to cancel the asynchronous operation. A task that represents the asynchronous operation. The task result contains a read-only list of function descriptors that satisfy the request criteria.

Optional interface implements by an that allows callbacks from the .

Allows a responder to augment a that the broker will send to the Copilot service. May be invoked multiple times for a given request.

Arguments containing relevant information available to the callback. Cancellation token. An augmented set of callbackArgs for the broker to use, or null to leave the request unchanged.

Optional interface for indicating model support inside of a responder provider.

Gets the models available through this responder.

The upstream models available through the Copilot service for this responder. Cancellation token. The models that this responder supports. Usually this will be the full set or a subset of the models available through the Copilot service (via ), but the responder may choose to surface an entirely different set of models if it has its own mechanism for issuing requests to those models.

Brokered service that provides responders per session.

Gets the identity of the responder.

Cancellation token. The identity of the responder. For agents, the identity returned here must match that which is provided to .

Create a responder for a specific session.

The id of the session to create the session for. Cancellation token. A new responder for this session. The lifetime of the responder can vary depending on VS lifetime, rpc connection, etc. Implementers should be prepared to fully restore any state required whenever this is called.

A is a wrapper around any MessagePack-serializable object which defers deserialization until requested. This allows serialized objects to be passed around in a type-agnostic way without the need to deserialize them until they are actually needed. This enables extensions to use custom types in Copilot that can be safely serialized without Copilot needing access to the assembly that defines the type. The types can be serialized in the extension before being sent to Copilot and then deserialized back in the extension after retrieving from Copilot.

To create a serialized object out of an arbitrary MessagePack-serializable object, simply call ."/>


             var serializedObject = SerializedObject.Create(new MyCustomType());

The inner object you provide is serialized into a raw byte stream inside the , allowing the to be safely passed across process boundaries or stored to disk. To deserialize the object, call . Deserialization will only succeed if the original type of the serialized object matches the requested type T.


             if (serializedObject.TryGetValue(out MyCustomType value))
             {
                 // use value here
             }

If you need to deserialize into a different type than was originally serialized (but which is still compatible with the original MessagePack schema), you can call which will attempt to deserialize the object into any type you request, regardless of the original type. This is useful in some backwards compatibility scenarios to attempt deserialization into a new type when the original type is no longer available or had breaking changes. Objects stored inside a should not be mutated after the is created since the exact timing of the serialization (and thus the snapshot of object contents) is not controlled by the caller.

Initializes a new instance of the class.

Assembly qualified type name of the data type stored. A sequence of bytes in MessagePack format. A value indicating whether the serialized object is JSON.

Gets the type name of the serialized object.

Gets a sequence of bytes in MessagePack format.

Gets a value indicating whether the serialized object is JSON.

Gets the length in bytes of the fragment.

Creates a new instance of from a value.

Type of the value. Copilot context value to serialize, this value must be serializable by MessagePack. an instance containing serialized data.

Creates a new instance of using GetType to get the target type.

Copilot context value to serialize, this value must be serializable by MessagePack. An instance containing serialized data. is .

Creates a new instance of with the given .

Type of the value. Copilot context value to serialize, this value must be serializable by MessagePack. An instance containing serialized data.

Creates a new instance of from a JSON string.

Assembly qualified type name of the data type stored. The JSON string representing the serialized object. An instance of containing the serialized data.

Tries to get the serialized value as the provided type if stored type is assignable to T.

Type to check for. Returns true if successful, false otherwise.

Tries to deserialize stored data to passed in type.

Deserialized value. Type to deserialize to. Returns true if successful, false otherwise.

Tries to deserialize stored data to passed in type.

Type to deserialize to. Deserialized value. Returns true if successful, false otherwise.

Checks if stored type is of the provided type T.

Type to check against. True if stored type information matches T, false if not or if stored type can not be loaded in given process.

Describes a subscription for an active interaction.

The ID of the interaction being tracked. The streaming interface.

Describes a subscription for an active interaction.

The ID of the interaction being tracked. The streaming interface.

The ID of the interaction being tracked.

The streaming interface.

Creates a progress content part and returns an to update it.

The response builder Cancellation token. A wired to report updates to the created content part. If is . The created will start in the status. Note that this is content progress, the final state of this progress will continue to be visible and persisted.

Gets an for an existing progress part.

The response builder. The existing progress part. Cancellation token. A representing the result of the asynchronous operation.

Reports a copilot progress update.

Progress. State message. Current status.

The progressive interaction between UI and the message pipeline.

Implement and provide this type to receive progress indications about steps, approve or deny side-effect steps, provide streaming content sinks and enable post processing steps for individual blocks.

Update a content part's current state.

Thrown if the type of doesn't match the original type.

Gets the current content part for the given id.

The id of the part to retrieve. Cancellation token. The content part with the . Thrown if the part id is not found. If the part is the currently streaming part, a snapshot of the part will be taken.

Updates the context used in the interaction to reflect context used in this interaction step.

The new list of context for the interaction. Cancellation token.

Represents a response that is still in flight.

While represents a fully completed response, is the sink for events (such as streaming content) from an in-flight response. For each final response that a responder begins to generate, will offer its subscribers the chance to provide a . Generally this ends up being the UI providing its streaming UI implementation.

Finishes the response with the given .

The reason that this response is finished. A cancellation token.

Builds responses from streamed content.

Finishes building the response and returns the completed result.

The status of the final response. A cancellation token. The completed response.

Represents a place for content to be streamed to.

Begins a new streamed block.

A description of the type of content that will be received in this block. A cancellation token. Thrown if called while there is already an open block if block nesting is not supported.

Writes content to the current block.

Any amount of content to be added to the current block as-is. A cancellation token. Thrown if there is no current block.

Ends the current block.

Thrown when called without a current block.

Writes a whole content part.

The content part to write to the stream. A cancellation token. Thrown when called while a block is open.

Redacts the entire content received by this sink.

A streaming content block of code.

Initializes a new instance of the class.

The id of the content block.

Gets the language associated with this code block.

Gets the file associated with this code block.

Base class for content block metadata.

The subclasses of this class are largely to denote metadata for its contents.

Initializes a new instance of the class.

The id of the content block.

Gets the content part id for this streaming block.

A streaming content block of text.

This is generally Markdown text content.

Initializes a new instance of the class.

The id of the content block.

Provides common trait keys used within the Copilot trait system.

Represents the common key used for the Language Version trait.

Represents the common key used for the Project Type trait.

Represents the common key used for the Test Frameworks trait.

Represents common Copilot type names used for traits in the Copilot system.

Represents the type name for the Language Version trait. Traits using this type should use a value.

Represents the type name for the Test Frameworks trait. Traits using this type should use a value.

Represents a trait for language versions associated with a specific programming language.

The programming language associated with the version. The version of the programming language.

Represents a trait for language versions associated with a specific programming language.

The programming language associated with the version. The version of the programming language.

The programming language associated with the version.

The version of the programming language.

Represents a trait containing information about the test frameworks in the context. It will also contain the programming language associated with the test frameworks.

The programming language associated with the test frameworks. A collection of test frameworks for the specified language.

Represents a trait containing information about the test frameworks in the context. It will also contain the programming language associated with the test frameworks.

The programming language associated with the test frameworks. A collection of test frameworks for the specified language.

The programming language associated with the test frameworks.

A collection of test frameworks for the specified language.

Indicates that an API is experimental and it may change in the future.

This attribute allows call sites to be flagged with a diagnostic that indicates that an experimental feature is used. Authors can use this attribute to ship preview features in their assemblies.

Initializes a new instance of the class, specifying the ID that the compiler will use when reporting a use of the API the attribute applies to.

The ID that the compiler will use when reporting a use of the API the attribute applies to.

Gets the ID that the compiler will use when reporting a use of the API the attribute applies to.

The unique diagnostic ID. The diagnostic ID is shown in build output for warnings and errors. This property represents the unique ID that can be used to suppress the warnings or errors, if needed.

Gets or sets the URL for corresponding documentation. The API accepts a format string instead of an actual URL, creating a generic URL that includes the diagnostic ID.

The format string that represents a URL to corresponding documentation. An example format string is https://contoso.com/obsoletion-warnings/{0}.

Specifies that null is allowed as an input even if the corresponding type disallows it.

Indicates that the specified method parameter expects a constant.

This can be used to inform tooling that a constant should be used as an argument for the annotated parameter.

Indicates the minimum bound of the expected constant, inclusive.

Indicates the maximum bound of the expected constant, inclusive.

Specifies that null is disallowed as an input even if the corresponding type allows it.

Applied to a method that will never return under any circumstance.

Specifies that the method will not return if the associated Boolean parameter is passed the specified value.

Initializes the attribute with the specified parameter value.

The condition parameter value. Code after the method will be considered unreachable by diagnostics if the argument to the associated parameter matches this value.

Gets the condition parameter value.

Specifies that an output may be null even if the corresponding type disallows it.

Specifies that when a method returns , the parameter may be null even if the corresponding type disallows it.

Initializes the attribute with the specified return value condition.

The return value condition. If the method returns this value, the associated parameter may be null.

Gets the return value condition.

Specifies that the method or property will ensure that the listed field and property members have not-null values.

Initializes the attribute with a field or property member.

The field or property member that is promised to be not-null.

Initializes the attribute with the list of field and property members.

The list of field and property members that are promised to be not-null.

Gets field or property member names.

Specifies that the method or property will ensure that the listed field and property members have not-null values when returning with the specified return value condition.

Initializes the attribute with the specified return value condition and a field or property member.

The return value condition. If the method returns this value, the associated parameter will not be null. The field or property member that is promised to be not-null.

Initializes the attribute with the specified return value condition and list of field and property members.

The return value condition. If the method returns this value, the associated parameter will not be null. The list of field and property members that are promised to be not-null.

Gets the return value condition.

Gets field or property member names.

Specifies that an output will not be null even if the corresponding type allows it. Specifies that an input argument was not null when the call returns.

Specifies that the output will be non-null if the named parameter is non-null.

Initializes the attribute with the associated parameter name.

The associated parameter name. The output will be non-null if the argument to the parameter specified is non-null.

Gets the associated parameter name.

Specifies that when a method returns , the parameter will not be null even if the corresponding type allows it.

Initializes the attribute with the specified return value condition.

The return value condition. If the method returns this value, the associated parameter will not be null.

Gets the return value condition.

Specifies that this constructor sets all required members for the current type, and callers do not need to set any required members themselves.

Specifies the syntax used in a string.

Initializes the with the identifier of the syntax used.

The syntax identifier.

Initializes the with the identifier of the syntax used.

The syntax identifier. Optional arguments associated with the specific syntax employed.

Gets the identifier of the syntax used.

Optional arguments associated with the specific syntax employed.

The syntax identifier for strings containing composite formats for string formatting.

The syntax identifier for strings containing date format specifiers.

The syntax identifier for strings containing date and time format specifiers.

The syntax identifier for strings containing format specifiers.

The syntax identifier for strings containing JavaScript Object Notation (JSON).

The syntax identifier for strings containing numeric format specifiers.

The syntax identifier for strings containing regular expressions.

The syntax identifier for strings containing time format specifiers.

The syntax identifier for strings containing format specifiers.

The syntax identifier for strings containing URIs.

The syntax identifier for strings containing XML.

Used to indicate a byref escapes and is not scoped.

There are several cases where the C# compiler treats a as implicitly - where the compiler does not allow the to escape the method. For example: for instance methods. parameters that refer to types. parameters. This attribute is used in those instances where the should be allowed to escape. Applying this attribute, in any form, has impact on consumers of the applicable API. It is necessary for API authors to understand the lifetime implications of applying this attribute and how it may impact their users.

A class that represents nullability info.

The of the member or generic parameter to which this NullabilityInfo belongs.

The nullability read state of the member.

The nullability write state of the member.

If the member type is an array, gives the of the elements of the array, null otherwise.

If the member type is a generic type, gives the array of for each type parameter.

An enum that represents nullability state.

Nullability context not enabled (oblivious)

Non nullable value or reference type

Nullable value or reference type

Provides APIs for populating nullability information/context from reflection members: , , and .

Populates for the given . If the nullablePublicOnly feature is set for an assembly, like it does in .NET SDK, the private and/or internal member's nullability attributes are omitted, in this case the API will return NullabilityState.Unknown state.

The parameter which nullability info gets populated. If the parameterInfo parameter is null. .

The parameter which nullability info gets populated. If the propertyInfo parameter is null. .

The parameter which nullability info gets populated. If the eventInfo parameter is null. .

Populates for the given If the nullablePublicOnly feature is set for an assembly, like it does in .NET SDK, the private and/or internal member's nullability attributes are omitted, in this case the API will return NullabilityState.Unknown state.

The parameter which nullability info gets populated. If the fieldInfo parameter is null. .

Polyfills for System.Private.CoreLib internals.

Represent a type can be used to index a collection either from the start or the end.

Index is used by the C# compiler to support the new index syntax


            int[] someArray = new int[5] { 1, 2, 3, 4, 5 } ;
            int lastElement = someArray[^1]; // lastElement = 5

Construct an Index using a value and indicating if the index is from the start or from the end.

The index value. it has to be zero or positive number. Indicating if the index is from the start or from the end. If the Index constructed from the end, index value 1 means pointing at the last element and index value 0 means pointing at beyond last element.

Create an Index pointing at first element.

Create an Index pointing at beyond last element.

Create an Index from the start at the position indicated by the value.

The index value from the start.

Create an Index from the end at the position indicated by the value.

The index value from the end.

Returns the index value.

Indicates whether the index is from the start or the end.

Calculate the offset from the start using the giving collection length.

The length of the collection that the Index will be used with. length has to be a positive value For performance reason, we don't validate the input length parameter and the returned offset value against negative values. we don't validate either the returned offset is greater than the input length. It is expected Index will be used with collections which always have non negative length/count. If the returned offset is negative and then used to index a collection will get out of range exception which will be same affect as the validation.

Indicates whether the current Index object is equal to another object of the same type.

An object to compare with this object

Indicates whether the current Index object is equal to another Index object.

An object to compare with this object

Returns the hash code for this instance.

Converts integer number to an Index.

Converts the value of the current Index object to its equivalent string representation.

Represent a range has start and end indexes.

Range is used by the C# compiler to support the range syntax.


            int[] someArray = new int[5] { 1, 2, 3, 4, 5 };
            int[] subArray1 = someArray[0..2]; // { 1, 2 }
            int[] subArray2 = someArray[1..^0]; // { 2, 3, 4, 5 }

Represent the inclusive start index of the Range.

Represent the exclusive end index of the Range.

Construct a Range object using the start and end indexes.

Represent the inclusive start index of the range. Represent the exclusive end index of the range.

Indicates whether the current Range object is equal to another object of the same type.

An object to compare with this object

Indicates whether the current Range object is equal to another Range object.

An object to compare with this object

Returns the hash code for this instance.

Converts the value of the current Range object to its equivalent string representation.

Create a Range object starting from start index to the end of the collection.

Create a Range object starting from first element in the collection to the end Index.

Create a Range object starting from first element to the end.

Calculate the start offset and length of range object using a collection length.

The length of the collection that the range will be used with. length has to be a positive value. For performance reason, we don't validate the input length parameter against negative values. It is expected Range will be used with collections which always have non negative length/count. We validate the range is inside the length scope though.

An attribute that allows parameters to receive the expression of other parameters.

Initializes a new instance of the class.

The condition parameter value.

Gets the parameter name the expression is retrieved from.

Initialize the attribute to refer to the method on the type.

The type of the builder to use to construct the collection. The name of the method on the builder to use to construct the collection. must refer to a static method that accepts a single parameter of type and returns an instance of the collection being built containing a copy of the data from that span. In future releases of .NET, additional patterns may be supported.

Gets the type of the builder to use to construct the collection.

Gets the name of the method on the builder to use to construct the collection.

This should match the metadata name of the target method. For example, this might be ".ctor" if targeting the type's constructor.

Indicates that compiler support for a particular feature is required for the location where this attribute is applied.

Creates a new instance of the type.

The name of the feature to indicate.

The name of the compiler feature.

If true, the compiler can choose to allow access to the location where this attribute is applied if it does not understand .

The used for the ref structs C# feature.

The used for the required members C# feature.

Indicates which arguments to a method involving an interpolated string handler should be passed to that handler.

Initializes a new instance of the class.

The name of the argument that should be passed to the handler. may be used as the name of the receiver in an instance method.

Initializes a new instance of the class.

The names of the arguments that should be passed to the handler. may be used as the name of the receiver in an instance method.

Gets the names of the arguments that should be passed to the handler.

may be used as the name of the receiver in an instance method.

Indicates the attributed type is to be used as an interpolated string handler.

Reserved to be used by the compiler for tracking metadata. This class should not be used by developers in source code.

Used to indicate to the compiler that a method should be called in its containing module's initializer.

When one or more valid methods with this attribute are found in a compilation, the compiler will emit a module initializer which calls each of the attributed methods. Certain requirements are imposed on any method targeted with this attribute: - The method must be `static`. - The method must be an ordinary member method, as opposed to a property accessor, constructor, local function, etc. - The method must be parameterless. - The method must return `void`. - The method must not be generic or be contained in a generic type. - The method's effective accessibility must be `internal` or `public`. The specification for module initializers in the .NET runtime can be found here: https://github.com/dotnet/runtime/blob/main/docs/design/specs/Ecma-335-Augments.md#module-initializer

Specifies the priority of a member in overload resolution. When unspecified, the default priority is 0.

Initializes a new instance of the class.

The priority of the attributed member. Higher numbers are prioritized, lower numbers are deprioritized. 0 is the default if no attribute is present.

The priority of the member.

Indicates that a method will allow a variable number of arguments in its invocation.

Specifies that a type has required members or that a member is required.

Reserved for use by a compiler for tracking metadata. This attribute should not be used by developers in source code.

Used to indicate to the compiler that the .locals init flag should not be set in method headers.

Initializes a new instance of the class.

Initializes a new instance of the class with the specified message.

An optional message associated with this attribute instance.

Returns the optional message associated with this attribute instance.

Returns the optional URL associated with this attribute instance.

Defines the context in which a JSON schema within a type graph is being generated.

The path to the schema document currently being generated.

The for the type being processed.

The declaring type of the property or parameter being processed.

The type info for the polymorphic base type if generated as a derived type.

The if the schema is being generated for a property.

The if a constructor parameter has been associated with the accompanying .

The corresponding to the property or field being processed.

Checks if the type, property, or parameter has the specified attribute applied.

The type of the attribute to resolve. Whether to look up the hierarchy chain for the inherited custom attribute. True if the attribute is defined by the current context.

Checks if the type, property, or parameter has the specified attribute applied.

The type of the attribute to resolve. Whether to look up the hierarchy chain for the inherited custom attribute. The first attribute resolved from the current context, or null.

Resolves any custom attributes that might have been applied to the type, property, or parameter.

The attribute type to resolve. Whether to look up the hierarchy chain for the inherited custom attribute. An enumerable of all custom attributes defined by the context.

Maps .NET types to JSON schema objects using contract metadata from instances.

The JSON schema draft version used by the generated schemas.

Generates a JSON schema corresponding to the contract metadata of the specified type.

The options instance from which to resolve the contract metadata. The root type for which to generate the JSON schema. The configuration object controlling the schema generation. A new instance defining the JSON schema for . One of the specified parameters is . The parameter contains unsupported configuration.

Generates a JSON object schema with properties corresponding to the specified method parameters.

The options instance from which to resolve the contract metadata. The method from whose parameters to generate the JSON schema. The configuration object controlling the schema generation. A new instance defining the JSON schema for . One of the specified parameters is . The parameter contains unsupported configuration.

Generates a JSON schema corresponding to the specified contract metadata.

The contract metadata for which to generate the schema. The configuration object controlling the schema generation. A new instance defining the JSON schema for . One of the specified parameters is . The parameter contains unsupported configuration.

Renders the specified instance as a JSON string.

The node to serialize. Whether to indent the resultant JSON text. The JSON node rendered as a JSON string.

Post-processing to enforce: - Every object with 'properties' has a 'required' array listing all property names. - Every such object has 'additionalProperties': false if not already specified. Works recursively for objects inside arrays (items), tuple-style items arrays, and inside properties of other objects. Note: This is required because we set "strict" mode for models using /responses API (eg. gpt-5-codex).

Registers the current schema node generation context; if it has already been generated return a JSON pointer to its location.

Controls the behavior of the class.

Gets the default configuration object used by .

Determines whether the '$schema' property should be included in the root schema document.

Defaults to true.

Determines whether the should be resolved for types and properties.

Defaults to true.

Specifies whether the type keyword should be included in enum type schemas.

Defaults to false.

Determines whether non-nullable schemas should be generated for null oblivious reference types.

Defaults to . Due to restrictions in the run-time representation of nullable reference types most occurrences are null oblivious and are treated as nullable by the serializer. A notable exception to that rule are nullability annotations of field, property and constructor parameters which are represented in the contract metadata.

Defines a callback that is invoked for every schema that is generated within the type graph.

ESgPGjGvZmP4gsH0khqssm2WWEc5HT9nmYltopnz7H8=HfScHnX3OslnPGojEbOwWy4G6IsCU3ezE1SjW2T7ixh1kbeBDxDqITezJkHD6Rhr3LIPO+cD6wOYj8UNjpnAflwSSJUBNxWBYE6tY2st3GG2fNgPKBVXmdINzgbdCevEYhvz9oE/MyaQCuxwN7OJdtO1IWm0lXCq29kqO0RmrdUy5iyRoBNYq9GD6SyA64s4o6J7EZQOvQSqV9HwEJQviFMk9NAM7cHXvoDlDlPjJtnaqS0tmkwZ+PRNz66K8AEKzerxmMU6M8IDLOauW8B0HsByPzUzTSrDcfKROfOU59m4mIYVezENo3p1KpAZwBhshY/3Ma16Yu8PPc1piWD8NA==1bGX4DipjN9Rz36FjqDRIsNEpQoiMVDAtCPTTFm7nCjsP3vZT6AKHoUFbukhuuVeBD862LJwZxTzaIuPx6DnY4c9apKxLeCOrRHMV1OqDnmPcxr3gv94gXroS2MTNzPz5HFKHmxfjXnZ5vDpHUj6A7vIplYhz0Kv/AkFLtFkUeKxPnTEX66Van5jYtqlgl/eE+DLHqYoxlZMBP/7SYNK8gImHR09+C0p5Rv0UgWZkERlmeYPI6pyo0T2q0qjH7dYL47lE1YLVjWX4HCxUiuVmtJsq6vDj3IExhrEYLp/rZ0kviMQ08VbADx9Ts7z48KJoLgcoVHvznL1DdA+Vpqe8Q==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