chat-arch-thrash-detect
Skill0 starsChanged 44 days agoMITView on GitHub
Calibration-gated PostToolUse hook that watches the rolling window of tool calls in a Claude Code session and emits a stderr nudge when it detects an Edit-thrash, Read-loop, Test-loop, or Tool-flail pattern. NOT a user-invokable skill — the directory exists to host the hook script + install/uninstall machinery alongside the rest of the chat-arch skill family. Gated behind `CHATARCH_THRASH_DETECT=1` for the 4-week calibration window (plan §Phase 4 #8).
In this repository15 other files it gives its agents.
- chat-history-searchSkill
- email-editorial-passSkill
- fable-seatSkill
- gemini-imageSkill
- global-review-loopSkill
- monitor-agent-threadSkill
- pattern-retrospectiveSkill
- review-loopSkill
- seo-index-validationSkill
- session-endSkill
- session-handoffSkill
- session-portalSkill
- ship-itSkill
- transcript-analysisSkill
- weekly-work-logSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 22
If you reached this skill via a slash command by mistake, no-op: there is nothing to invoke here.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
chat-arch-thrash-detect
This skill is a holder for the PostToolUse hook at
post-tool-use-hook.cjs. There is no slash-command surface — the hook
runs automatically on every tool call when the user has run
install.cjs AND set CHATARCH_THRASH_DETECT=1 in their environment.
If you reached this skill via a slash command by mistake, no-op: there is nothing to invoke here.
Calibration
Threshold values are duplicated in thresholds.cjs to keep the hook
hermetic (no chat-arch package import). The canonical source is
packages/analysis/src/thresholds.ts in the chat-arch repo
(THRESHOLDS.thrash). When that file changes, mirror the change
here.
Fire events are appended to
~/.claude/cache/chat-arch-thrash/fires.jsonl so the calibration pass
can audit fire rate, false-positive rate, and ack-or-pivot follow-up
rate per the plan's pre-registered launch criteria.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.