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

compass-evaluator

Skill2 starsChanged 32 days agoMITView on GitHub

Runs bounded evaluation experiments comparing technical alternatives

The file

We split this skill into the rules it gives an agent

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

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    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:

    If a disagreement is high-cost or security-sensitive, escalate with alternatives.

    line 38
  2. Distinctive:

    Do not mutate matcher weights or Skill sidecars from an evaluation

    line 53
  3. Very specific:

    Do not treat evaluation winners as IMPLEMENTATION_PLAN approval

    line 54
  4. Distinctive:

    Do not call live network Technology Intelligence as part of M3 evaluations

    line 55

≈ “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

Compass Evaluator

Use this Skill when

The Captain or First Mate needs a bounded comparison of alternatives (A/B), architecture options, or experiment spikes — without expanding into product implementation.

Inputs

  • Plan ID
  • Objective / question under evaluation
  • At least two alternatives (id + label)
  • Recommendation text
  • Optional winner alternative id, hypothesis, evidence paths

Procedure

  1. Confirm the evaluation is informational — it does not approve product edits.

  2. Record the experiment:

    ./scripts/run-evaluation.sh \
  --plan-id <plan-id> \
  --objective "<question>" \
  --alternatives '[{"id":"a","label":"Option A"},{"id":"b","label":"Option B"}]' \
  --recommendation "<recommendation>" \
  --winner a
  3. Store evidence under .agent/evidence/<plan-id>/ when applicable.
  4. Present the evaluation id and recommendation to the Captain.
  5. If a disagreement is high-cost or security-sensitive, escalate with alternatives.

Companion subagent

Cursor subagent compass-evaluator may be invoked for the same procedure when the First Mate wants a dedicated evaluation pass. Proficiency for this agent is tracked separately via Skill experience-routing / record-agent-proficiency.sh with Captain-approved metadata.

Output

Schema-valid Evaluation JSON under .agent/evaluations/ (gitignored runtime files).

Prohibited actions

  • Do not mutate matcher weights or Skill sidecars from an evaluation
  • Do not treat evaluation winners as IMPLEMENTATION_PLAN approval
  • Do not call live network Technology Intelligence as part of M3 evaluations

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