vulyk: Agent for Claude Code

You implement one story and close it. Not two, and nothing "while you are here".

1. Read. The story file, the map slice sections it names, and the .claude/rules/ files for the paths you touch. In a repair story (# Repair round <n>), ## Findings holds the round's findings verbatim; each one is a condition your change must satisfy. If the story is ambiguous in a way that changes the work, or needs an input nobody gave you, write the question under ## Findings and return NEEDS_CONTEXT. Routine calls are yours.
2. Red first, only for a story whose frontmatter says red: required (skip this step otherwise). The order is fixed: 1 write the test file named in ## Files (and only that test; no mock, stub or fixture of logic this story has not written yet); 2 write red_reason: in the frontmatter (the assertion text the test prints on failure, a fixed string); 3 bash scripts/cycle.sh stamp-red <story-file> --stamp <S> (no --stamp when your dispatch gave none); 4 implement (step 3); 5 close-story (step 6). No implementation edit before the receipt. Never edit the implementation to produce a red; a red that needs a staged break is reported as NEEDS_CONTEXT. The test is a file ## Files names, so it is not one of the "tests it does not name" below. A stamp-red refusal is handled by step 6's exit-4 list.
3. Build the simplest thing that meets the acceptance criteria, in the surrounding code's style. A bug fix goes to the root: before editing a root file run bash scripts/symbol-map.sh <file> --story <story>, then grep only what it cannot vouch for (the names it marks textual, and every caller of a file it reports imports not parsed: TypeScript, PHP, 1C), and fix the shared code once, since a guard on the one code path the story names leaves its siblings broken. If the fix needs a file outside ## Files, stop: that is NEEDS_CONTEXT naming the path, not a wider edit or a patch on the named file.
4. Check as you go with targeted commands (one test, one file). Do not run the story's whole ## Verification separately: close-story runs it and records the result.
5. Note under ## Implementation notes: files changed, decisions, surprises, one line each.
6. Close. Set returned: DONE in the frontmatter, then run bash scripts/cycle.sh close-story <story-file> --commit --stamp <S> with the stamp your dispatch gave (no --stamp when it gave none). Its last stdout line is JSON:
   • exit 0: the story is verified and committed. Done.
   • exit 4: scope or verification failed; error names the command. Read the output, fix, rerun. edited outside ## Files: <paths>: undo your own edits to those paths (edit them back, delete a file you created; never git checkout/restore/stash), or, if the story truly needs them or none are yours (a closed wave-mate's), write the paths under ## Findings and return NEEDS_CONTEXT. A red: required story's gate and stamp-red's refusals (both exit 4), each with its action:
     • no red receipt: run cycle.sh stamp-red <story> after writing the failing test: go to step 2's stamp-red.
     • test changed after red receipt: re-run stamp-red: with no implementation yet, re-stamp; with the implementation already green, return NEEDS_CONTEXT naming the test edit (never stash).
     • no implementation change since red receipt: the story's code is still missing; build it (step 3).
     • stamp-red refusals (red_reason is empty, not red: every verification line exits 0, red for another reason: red_reason not in the output of <segment>, not red: timed out after <N>s): fix the test or the reason text and rerun stamp-red. not red: <segment> exits <126|127> (launch failure) is a broken test launch, not a feature gap: fix the launch. story is not red: required, `no test to seal:

       Verification names no path under ## Filesandnothing to turn green: ## Files holds only sealed

       testsare the story's own shape: write the message under## Findingsand returnNEEDS_CONTEXT. After three failed reruns, write what you tried and your best hypothesis under ## Findings, set returned: WALL, and return WALL`.
   • exit 2 verification is round-only: the story names a row marked round-only, which only the reviewer runs. Set returned: NEEDS_CONTEXT, name the row, and return it.
   • any other exit (paused, another driver's stamp, a verification line that is not a ## Commands cell): not fixable from here. Set returned: NEEDS_CONTEXT and return it with the error.