sanitize
Skill0 starsChanged 58 days agoMITView on GitHub
Close out completed work with scoped commit/push and Canon updates when the user says "收尾", "提交并保存", or "commit and save". Generate a formal evidence-backed task report on "完整任务技术报告", "成果汇报", or "post-task report". Report mode does not commit or push without closeout intent.
2 other projects use a version of this.
Adopted by
Most recently taken up by
- louievandyke/sanitize-clitheir own version, first seen Oct 20, 2026
- Stackbilt-dev/aegis-osstheir own version, first seen Oct 8, 2026
Other versions2 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 2
- Stackbilt-dev/aegis-oss★ 3same name
- louievandyke/sanitize-cli★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository22 other files it gives its agents.
- AGENTS.mdAGENTS.md
- acquisitionSkill
- after-actionSkill
- breachSkill
- cartographic-reconSkill
- chartSkill
- codifySkill
- conopsSkill
- coverSkill
- dossierSkill
- executeSkill
- go-nogoSkill
- herdrSkill
- netease-mailSkill
- neutralizeSkill
- paperworkSkill
- passdownSkill
- repairSkill
- taskingSkill
- tracebackSkill
- warroomSkill
- x-likes-digestSkill
The file
We split this skill into the rules it gives an agent
12 rules. About 0 are common, 9 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈9
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- could be automatic: a setting could enforce it
- Distinctive: line 11
Read the shared Canon contract before finalizing meaningful work:
- Distinctive: line 43
Use the shared Review Gate only when equivalent evidence is not already recorded by execute, tasking, repair, or traceback.
- Distinctive: line 45
Use ordinary scoped Git for local commit/push.
- Very specific: line 46
If the requested outcome includes a GitHub PR, hand off publication to $github:yeet instead of duplicating PR creation logic here.
- Very specific: line 51
Read references/formal-report.md completely, then:
- Distinctive: line 56
Write the 6-section Markdown report to the user-specified or repo-appropriate artifact path.
- Distinctive: line 67
If the task is still active, generate an interim report only when explicitly requested and label missing validation.
- Distinctive: line 86
Never use git add . or commit unrelated dirty work.
- Distinctive: line 87
Do not revert, reset, or overwrite changes you did not make unless the user explicitly asks.
- Distinctive: line 88
Do not commit/push when Review Gate is blocked.
- Distinctive: line 90
If no Canon task page can be resolved, stop and ask or create one through the documented task resolution path before claiming wrapup is …
- Very specific: line 109
Use route=closeout|report|both|unresolved.
≈ “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
Completion Closeout
完成开发后的 scoped publish、Canon closeout,或 formal task report。
Read the shared Canon contract before finalizing meaningful work:
/home/yhr/.agents/repos/agent-skills/references/canon-output-contract.mdMode Routing
- Closeout — user asks to commit, push, save, or wrap up completed work.
- Report — user asks for a formal end-to-end report on completed work. Read
references/formal-report.md; do not enter git mutation steps. - Both — user explicitly requests report plus closeout. Generate and gate the report first, include it in the confirmed file scope, then publish and promote.
A one-incident troubleshooting story belongs to after-action. One reusable wrong/correct/trigger guardrail belongs to codify.
Closeout Mode
1. 检查 git 状态
2. 识别当前任务相关文件
3. 生成并确认 commit message
4. 执行 Review Gate(如本 diff 尚未通过)
5. 执行 git commit + push
6. 更新 Canon task page(替代旧的 runtime conversation 保存)
7. 更新 Canon task page § Progress / § Evidence / § Timeline
8. 写入 Canon update card / artifact referenceReuse runtime-native review and Git capabilities:
- Use the shared Review Gate only when equivalent evidence is not already recorded by execute, tasking, repair, or traceback.
- Use ordinary scoped Git for local commit/push.
- If the requested outcome includes a GitHub PR, hand off publication to
$github:yeetinstead of duplicating PR creation logic here.
Formal Report Mode
Read references/formal-report.md completely, then:
- Resolve the completed Canon task page.
- Gather implementation and validation evidence from linked artifacts, commits, tests, planning records, and the current repository.
- Write the 6-section Markdown report to the user-specified or repo-appropriate artifact path.
Run:
python3 <skill-dir>/scripts/report_gate.py <report.md>- Record the report's absolute path and durable conclusions in the Canon task page or an update card.
If the task is still active, generate an interim report only when explicitly requested and label missing validation. Report mode alone never commits, pushes, or marks the task done.
Workflow Gate Contract
sanitize is the final commit/push/Canon promotion gate and follows the shared workflow output contract:
/home/yhr/.agents/repos/agent-skills/references/skill-output-contract.mdCloseout consumes prior implementation, validation, Review Gate, and traceback evidence. Report mode consumes the same evidence but does not publish code. Neither mode should create new product behavior while completing the task.
Gotchas
- Never use
git add .or commit unrelated dirty work. Scope files from the Canon task page, artifacts, and current diff. - Do not revert, reset, or overwrite changes you did not make unless the user explicitly asks.
- Do not commit/push when Review Gate is blocked. Fix blockers or record explicit user waiver first.
- A commit without Canon task/update-card progress is incomplete for durable workflows. Record commit hash, review evidence, artifact refs, and next step.
- If no Canon task page can be resolved, stop and ask or create one through the documented task resolution path before claiming wrapup is complete.
Closeout execution details
For Closeout mode, read the scoped Git and Canon steps before any mutation.
边界情况
- 无改动文件:提示用户确认是否只做 Canon 收尾。
- commit message 生成失败:让用户手动输入。
- push 失败:处理冲突后重试,或提示用户手动 push。
- task page 不存在:按
references/canon-task-resolution.md创建新 task page。 - Canon 不可用:继续 git 收尾,并明确 Canon promotion 未完成。
- 未配置 git user:提示用户配置
git config user.name/email。
Routing telemetry
Emit one final local event following
the shared routing contract.
Use route=closeout|report|both|unresolved; set expected-route only from clear
user/caller intent. Record publish=none|commit|push for the highest completed
Git action, never the planned action. Report-only must have publish=none.
Record report, report-gate, review, commit, push, canon step states;
non-applicable steps are skipped, planned work is pending, failed work is
error or blocked. No publishable changes is skipped, not a new commit.
Attach report/task/gate references; no commit message or report text in telemetry.
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.

