agentleFS
arnaudgelasarnaudgelas/K-Ops/.claude/skills/qa-agent/SKILL.md

qa-agent

Skill1 starsChanged 7 months agoMITView on GitHub

Answer questions from the vault, then file durable insights back into the knowledge base.

9 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use release-manager.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • exzmSkill

    release-manager

    Manages releases as a Release Manager/Tech Lead. Verifies all steps are completed successfully, checks tests, code review status, documentation, and requirements compliance. Creates commit messages, release notes, determines version bumps, and makes release decisions. Use when preparing releases, creating changelogs, or when the user requests release management.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

10 rules. 0 are common, 1 is distinctive, and 9 are very specific to this project.

  • 0

    Common

    Most of the 9 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 9

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Read notes/Home.md or notes/Indexes/Flat_Concept_Index.md, and run search_vault.py programmatically.

    line 12
  2. Very specific:

    Must have a top-level heading (e.g.

    line 35
  3. Very specific:

    Must have a second-level heading (e.g.

    line 36
  4. Very specific:

    Must have a ## Vault Updates section.

    line 37
  5. Very specific:

    If answer_quality: durable, this section must list the exact edits made to concept pages.

    line 38
  6. Very specific:

    If answer_quality: memo-only, this section must be exactly - None..

    line 39
  7. Distinctive:

    Do not fabricate evidence.

    line 44
  8. Very specific:

    Every factual claim in the answer must carry an inline wikilink citation: [[Sources/<source_id>|<source_id>]] or [[Concepts/<name>|<name>]].

    line 45
  9. Very specific:

    If gap found: Add to the concept page's ## Open Questions and notes/TODO.md.

    line 52
  10. Very specific:

    If contradiction found: Add to notes/Maintenance/Contradictions.md.

    line 53

“Common” is measured: kept by at least 6 in 10 of the 9 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Q&A Agent

Goal

Produce useful answers without letting knowledge disappear into chat.

Workflow

  1. Read notes/Home.md or notes/Indexes/Flat_Concept_Index.md, and run search_vault.py programmatically.
  2. Classify the query using notes/Runbooks/Query_Planner.md (one of: lookup, synthesis, contradiction, freshness, code, audit, research).
  3. Execute the required retrieval layers in the order specified for that query class.
  4. Find the most relevant concept pages and source summaries (capped at reading at most 10 concept pages total).
  5. Answer from the vault.
  6. Save the answer memo in notes/Answers/YYYYMMDDTHHMMSS-topic.md using the schema below.
  7. File durable improvements back into concept pages, open questions, contradictions, or TODOs.

Answer Memo Schema

Frontmatter:

  • title (Human-readable topic title)
  • type: answer
  • asked_at (ISO-8601 date)
  • answer_quality (must be exactly durable if updating the vault, or memo-only if not)
  • scope (must be exactly shared for durable answers, or private for memo-only answers)
  • query_class (must be one of: lookup, synthesis, contradiction, freshness, code, audit, research)
  • sources_consulted (YAML list of every source_id opened during the session)
  • retrieval_path (YAML list of dictionaries specifying each retrieval step taken; each step must have keys: method, layer, query, results_count)
  • fetch_required (boolean, indicating whether external web fetching was performed/required)
  • tags (must include kb/answer)

Structure:

  • Must have a top-level heading (e.g. # Question / # Answer).
  • Must have a second-level heading (e.g. ## Summary, ## Analysis).
  • Must have a ## Vault Updates section.
    • If answer_quality: durable, this section must list the exact edits made to concept pages.
    • If answer_quality: memo-only, this section must be exactly - None..

Rules

  • Prefer vault-grounded answers.
  • State uncertainty clearly (e.g., "according to [[Sources/...]]" vs "this is unclear").
  • Do not fabricate evidence.
  • Every factual claim in the answer must carry an inline wikilink citation: [[Sources/<source_id>|<source_id>]] or [[Concepts/<name>|<name>]]. If no source supports a claim, mark it (unverified).
  • For a consequential answer, gate the evidence by stakes: run consequence-gate --tier <recommendation|decision|autonomous> (optionally --concept <stem>). Do not let blocked, quarantined, or unsupported claims back a high-consequence answer; flag the shortfall in the memo instead.

Write-Back Requirement

The Q&A skill is not complete until at least one write-back path is executed:

  1. Always: Save the answer memo under notes/Answers/ and ensure quality/scope/updates match.
  2. If new claim found: Append to the relevant concept page's ## Key Claims and ensure the concept has ≥ 90% direct citation coverage or demote it.
  3. If gap found: Add to the concept page's ## Open Questions and notes/TODO.md.
  4. If contradiction found: Add to notes/Maintenance/Contradictions.md.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.