Three stop-hook bugs that break loop termination:
- A bare assistant message equal to the promise (e.g. just "DONE") ended
the loop even without <promise> tags: the perl substitution passed the
input through unchanged when the tag was absent. Now the promise text
is only extracted when a <promise>...</promise> tag is present.
- The observed promise text was whitespace-normalized but the expected
promise was not, so a promise containing consecutive spaces
("ALL DONE") could never match and the loop never ended. Both sides
are now normalized identically (trim + collapse runs of whitespace).
- The iteration reader accepted "iteration:1" (no space after the colon)
but the writer's sed only matched "iteration: " with a space, so the
counter froze and the loop ran past max_iterations. The writer now
matches with or without spaces, like the reader.
Adds tests/stop-hook.test.sh covering all three regressions plus
controls; the three bug cases fail on the previous script and pass now.
Fixesanthropics/claude-code#81827Fixesanthropics/claude-code#81828Fixesanthropics/claude-code#81829
No-Verification-Needed: shell-script plugin in claude-plugins-official; no claude-cli-internal runtime surface
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- Split pipeline into two steps (extract lines, then parse) mirroring the
original structure.
- set +e around the jq call so failures reach the $? check instead of
aborting under set -e.
- The "no text content" branch remains removed (that was the original bug —
all-tool-use turns now correctly yield empty text and the loop continues).
The state file lives at .claude/ralph-loop.local.md — project-scoped,
not session-scoped. The plugin's Stop hook fires in every Claude Code
session open in that project directory. So if session A starts a loop,
session B's Stop events also find the state file and block, feeding A's
prompt into B and consuming A's iteration budget.
This was masked by the transcript-parsing bug fixed in the previous
commit: that bug deleted the state file on the first Stop in any
session, so neither session looped. Fixing it exposed the leak.
Fix: setup writes CLAUDE_CODE_SESSION_ID into the frontmatter; the hook
compares against .session_id from its stdin JSON and exits silently on
mismatch. State files without session_id (written by old setup scripts)
fall through to preserve existing behavior.
Claude Code writes each assistant content block (text/tool_use/thinking)
as its own JSONL line. The hook's `grep role:assistant | tail -1` would
grab whichever block happened to be last — often tool_use — then jq's
text filter returned empty string, triggering the 'no text content' path
which deletes the state file and exits without blocking.
Net effect: the loop silently never fires. In one observed session, 62%
of assistant lines were tool_use-only; the hook deleted state on the
very first Stop event every time.
Fix: slurp all assistant lines with jq -rs, flatten to text blocks only,
take the last. Empty result is now non-fatal — no text means no <promise>
tag, so the loop continues. Also absorbs jq parse errors (control chars
in text) via || fallback instead of aborting under set -e.
- Rename plugin from "ralph-wiggum" to "ralph-loop" to avoid trademark concerns
- Update all internal references to use "Ralph Loop" as the prominent name
- Keep explanatory text noting it "implements the Ralph Wiggum technique" (allowed)
- Rename plugin directory from plugins/ralph-wiggum to plugins/ralph-loop
- Update marketplace.json with new plugin name and source path
- Update plugin-dev documentation references
This change follows legal's recommendation to replace "Wiggum" with "Loop"
in the plugin name while still explaining the technique origin.
Slack thread: https://anthropic.slack.com/archives/C09KU300P7F/p1767741142753959
Co-authored-by: Claude <noreply@anthropic.com>
