agentleFS
Sign inStart free
loganware05loganware05/captains-compass-cursor/.cursor/skills/knowledge-steward/SKILL.md

knowledge-steward

Skill2 starsChanged 32 days agoMITView on GitHub

Ingests and queries the project knowledge store for planning context (explicit CLI)

  • Reads credentials

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Search query for readback

    line 18
  2. Very specific:

    When .agent/knowledge/vector-index.json exists, capability plans use hybrid search for Knowledge Context by default.

    line 50
  3. Distinctive:

    Use the Knowledge Context section in capability plans as informational input only — it does not alter Skill rankings or matcher weights.

    line 78

≈ “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.

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

Knowledge Steward

Use this Skill when

The Captain or First Mate needs durable project knowledge surfaced during planning — Experiences, evaluations, routing artifacts, execution runs, and ADR headings from DECISIONS.md.

Inputs

  • Explicit file paths or store roots (experience, evaluations, routing, runs, decisions, procedures, notion, notebooklm)
  • Search query for readback

Procedure

  1. Ingest (explicit CLI only — never auto-run on workstream close):

    ./scripts/ingest-knowledge.sh --from-store experience,evaluations,decisions
./scripts/ingest-knowledge.sh --paths tests/fixtures/experience/contact-counter.json

    decisions auto-ingests ADR headings from DECISIONS.md.

    For Captain-exported Notion / NotebookLM files, use Skill external-knowledge-ingest or:

    ./scripts/ingest-knowledge.sh --from-store notion,notebooklm

  2. Query for planning context (read-only):

    ./scripts/query-knowledge.sh --query "evaluator routing" --kind decision
./scripts/query-knowledge.sh --query "matcher tuning" --kind performance --mode hybrid
./scripts/ingest-knowledge.sh --from-store runs,experience

    Execution runs ingest as kind: performance with performance_metrics (outcome, retries, skills, agents, models). Re-ingest overwrites existing know-run-* / know-exp-* items idempotently.

    When .agent/knowledge/vector-index.json exists, capability plans use hybrid search for Knowledge Context by default. Override with COMPASS_KNOWLEDGE_SEARCH_MODE=keyword|vector|hybrid.

  3. Rebuild vector index (explicit CLI only):

    ./scripts/rebuild-knowledge-vector-index.sh
./scripts/ingest-knowledge.sh --from-store decisions --rebuild-vector

    For fixture dense embeddings (TF-IDF remains fallback), use Skill embedding-providers:

    COMPASS_EMBEDDING_PROVIDER=fixture \
  ./scripts/rebuild-knowledge-embedding-index.sh
  4. Review matches under .agent/knowledge/items/ with provenance.

  5. Optionally propose a reusable procedure (staging + Captain PR only):

    ./scripts/propose-procedure-from-knowledge.sh \
  --item-ids know-adr-020 \
  --title "Bounded weight apply playbook"
  6. Use the Knowledge Context section in capability plans as informational input only — it does not alter Skill rankings or matcher weights.
  7. Review Performance Context in capability plans (always rendered; empty when no kind: performance items match).
  8. For procedure playbooks, use Skill procedure-playbooks or --from-store procedures after staging/approved playbooks exist.

Output

  • Knowledge items under .agent/knowledge/items/
  • Rebuilt .agent/knowledge/index.json
  • Optional .agent/knowledge/vector-index.json after explicit rebuild
  • Optional .agent/knowledge/embedding-index.json after fixture embedding rebuild
  • Ingest audit under .agent/knowledge/ingest-log/
  • Optional procedure staging under .agent/knowledge/procedures/staging/

Prohibited actions

  • Auto-ingest on workstream close without explicit CLI
  • Silent install of procedures into .cursor/skills/ or rules
  • Mutating matcher weights from query results
  • Ingesting secret paths (.env, credentials)

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.

Check your setup, free