| name | home-history-distill |
| description | AIエージェント履歴を安全に集計し、再利用できる skills / agents / rules / scripts 候補だけを抽出する。履歴本文を公開せず、秘密情報を先に検出して抽象化したいときに使用。 |
home-history-distill
Past prompts are useful raw material, but they are not publishable artifacts.
Use this skill to turn local AI history into reusable dotfiles assets without
leaking conversation content.
Workflow
- Run the prompt-review collector into
/tmp, not the repository:
python3 .agents/skills/prompt-review/scripts/collect.py --days 0 > /tmp/dotfiles-prompt-review-data.json
- Inspect aggregate data first:
- total messages and tools
- top projects
secret_warnings- category counts
python3 scripts/summarize-ai-history.py /tmp/dotfiles-prompt-review-data.json
-
If secret_warnings is non-empty:
- do not quote the raw value
- tell the user which project/tool/timestamp contains a credential-like item
- recommend rotation or invalidation
- update detection patterns if the warning came from an uncaught format
-
Extract reusable patterns:
- repeated multi-step work -> skill
- specialized read-only judgment -> agent
- durable always-on convention ->
AGENTS.md or .agents/rules/
- deterministic local check -> script
-
Write only abstracted instructions. Do not include raw logs, credentials,
private endpoints, account IDs, customer names, or personal paths.
-
Validate:
git secrets --scan
./scripts/audit-agent-config.sh
Output
Report the data scope, high-frequency categories, created assets, skipped
candidates, and any secret-warning remediation needed.
