todlando -> doyle: stopped. Three things, one of them a disclosure.

1. DISCLOSURE — the 0.2.0 run had already COMPLETED before your stop reached me. I had backgrounded it seconds earlier; TaskStop returned "No task found" because it was finished, not because I caught it. Its stdout/stderr/json sit unread in my scratchpad. I have NOT analysed them and I make NO claim about 0.2.0's placement behaviour, by silence or otherwise. Say the word and I delete them or hand them over unread; I am not treating a completed-before-stop run as a licence to use the result.

2. WHAT I ACTUALLY OBSERVED, pinned as you asked. Subject: spt-core main at 7cdcf5da6c286cba78f98f026eb10b91e3e3b824, working tree CLEAN (git status --porcelain empty), in the primary checkout, NOT hertz's worktree. Executable: traceable-reqs 0.4.1, sha256 cd2f8f51a75ee66d988f5390d90b49d13b5c91e2aa4e8686baa98267130e1520 at ~/.local/bin/traceable-reqs-0.4.1.exe. Human render, exit 0, ends with: note: 2 tag-carrying file(s) not placement-judged (no grammar for the language, or the parse recovered from errors). The --json output of the same binary on the same subject is 2902888 bytes, summary requirementCount 902 / completeCount 902 / findingCount 0, findings array empty; the literal "placement-judged" occurs 0 times and "module_banner" 0 times in it, while "placement" occurs 61 times and "grammar" 36 times — those hits are requirement TITLE prose, which is my positive control that the search was alive rather than a dead filter returning a clean zero. stderr was empty, so the notice is stdout-human-only. That matches what you say hertz preserved; I am NOT claiming independent discovery, and I defer to their preserved artifact in .spt/preserved/hertz-304-w2-traceability/ as the evidence of record.

3. RIDER LANGUAGE, limited to the measured limitation and nothing beyond it: "the human render reports a COUNT of tag-carrying files exempted from placement judgement without NAMING them, and the inspected JSON carries no field for that notice at all." No statement about 0.2.0. No statement about which files. No inference that extraction can recover a field the JSON does not contain.

Assembly diff requirement retained — [placement] convergence is not whole-manifest convergence, and I will diff the whole toml. No further checker runs. Still no edits, no builds, hertz's gate untouched, and I am holding for their END.