qa-audit
Skill9 starsChanged 20 days agoMITView on GitHub
Run the full-surface QA health check of the three ask-marcel-office bins (auth, read and write) by executing docs/QA-PLAYBOOK.md phase by phase — offline gates, bundle smoke tests, manifest-driven parameter matrix, conversion contract matrix, live read-only Graph drift probes, agent-ergonomics journeys, and gap/roadmap analysis. Use when the user asks to audit the CLI, run QA, check CLI health, find bugs/regressions across the command surface, or re-verify after a Microsoft Graph change. Produces a severity-ranked report and a proposed fix plan; never fixes code during the run.
3 other projects use a version of this.
Adopted by
Most recently taken up by
- florian101010/awesome-agentic-AI-coding-templatetheir own version, first seen Nov 1, 2026
- Myriambs/argus-qatheir own version, first seen Nov 1, 2026
- ngreenwall/claude-skillstheir own version, first seen Jul 9, 2026
Other versions3 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 3
- florian101010/awesome-agentic-AI-coding-template★ 4same name
- ngreenwall/claude-skills★ 2same name
- Myriambs/argus-qa★ 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 repository7 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- atelier-greenfieldSkill
- atelier-grill-meSkill
- atelier-review-meSkill
- atelierSkill
- simplified-technical-englishSkill
- ask-marcelSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 0 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
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
- Very specific: line 12
Read-only tenant access: Only commands whose manifest effect is read (playbook §0 rule 1.
- Very specific: line 14
Compare against the previous run: Read the most recent .claude/qa-reports/*.md first.
- Very specific: line 16
Keep the playbook honest: If a check is wrong/stale (a contract changed intentionally, a fragile probe is obsolete), update …
≈ “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
QA audit runner
You are auditing the three bins ask-marcel-office-auth, ask-marcel-office-read and ask-marcel-office-write. The authoritative procedure is docs/QA-PLAYBOOK.md — read it in full before doing anything, then execute phases A→H in order. This skill only adds the operating rules:
- Track phases as tasks. One task per playbook phase (A–H); mark in_progress/completed as you go. Full surface every run — no sampling, no tiers.
- Report, then plan. You find and document; you do NOT fix during the run (not even trivial P1s — note them, finish the sweep). After delivering the report, propose a severity-ordered fix plan and WAIT for approval.
- Read-only tenant access. Only commands whose manifest
effectisread(playbook §0 rule 1; the mail drafts and the PDF conversions are the write commands ofask-marcel-office-write). Respect every safety rule in playbook §0 (no external URL fetches, single-quoteb!ids,--output-pathfor multi-MB bytes). - Raw findings are private. Write the full report to
.claude/qa-reports/YYYY-MM-DD.md(gitignored — tenant file names/subjects/IDs may appear there). The ONLY committed artifacts of a run are: the sanitized Run-log row appended to playbook §I, plus any updates to the §E2 drift register and §G roadmap register. - Compare against the previous run. Read the most recent
.claude/qa-reports/*.mdfirst; anything that passed before and fails now is a REGRESSION and outranks same-severity new findings. - Manifest-driven, not hand-typed. Regenerate worklists from
read/dist/commands.jsonandwrite/dist/commands.json(snippets are in the playbook) so the audit scales as commands are added. - Keep the playbook honest. If a check is wrong/stale (a contract changed intentionally, a fragile probe is obsolete), update
docs/QA-PLAYBOOK.mdin the same run — the playbook is a living document, but contract-matrix changes need the maintainer's explicit OK. - Budget: a full run is several hours of focused work. If context runs low mid-run, write partial findings to the report file FIRST, then continue — never lose evidence.
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.