PERRI'S #300 FEASIBILITY MESSAGES — PRESERVED VERBATIM AS SOURCE EVIDENCE. Preserved on doyle's instruction (3DDGBBZE, 2026-09-12): the contract's summary of these is not a substitute for the source. docs/INPUT-PROVENANCE-CONTRACT.md §7 paraphrases them; THESE are the bytes. perri-4JZ4L5RC-claude-spt-feasibility-answer.txt From: perri (claude-spt adapter owner). Received by todlando on HFENDULEAM 2026-09-12T20:26:58Z. Answers the #300 feasibility request against the SUBMITTER IDENTITY frame. Withdraws perri's earlier co-authorship objection. Carries the MEASURED enqueue result (n=3, hook at +0.5/+0.9/+1.0s after Enter, one fire per submission, no second fire) WITH its three caveats: PTY-typed rather than human-typed, endpoint forced idle to take that route, and the stop-to-next-turn window unmeasured and asserted in neither direction. Also reports machine injection indistinguishable at the harness level (their own probe was one), edit/split undetectable, and the hook's ~5s deadline with p99 4835ms. perri-AKMIT2KO-population-precision-fix.txt From: perri. Received 2026-09-12T20:28:26Z, unprompted, correcting MY readback of their answer. I had merged two separate populations into one sentence. The corrected figures: ALL submissions 302 audited / 227 hook-covered, the 75 misses being harness-synthesised transcript entries or built-in local slash commands; PTY-TYPED submissions 133 audited / 103 covered, where all 30 misses are safe-mode sessions with hooks off BY DESIGN. Two different holes needing different contract language. Also carries the instrument limit: the audit observes hook OUTPUT in the transcript, not hook INVOCATION, so a hook that fired and emitted nothing is invisible to the count; the coverage figures are their own positive controls. DELIVERY NOTE, so the file shape is not mistaken for tampering: both arrived as sptc spill files because each exceeded the harness's inline context cap. The envelope and the SPT-NOW-SIGNAL block are the delivery wrapper as received; the message body inside is unmodified.