agentleFS
fqmyysjjdfqmyysjjd/agent-feed/.agents/skills/guidance-promoter/SKILL.md

guidance-promoter

Skill4 starsChanged 6 months agoMITView on GitHub

Use when user corrections, repeated AI failures, or session-state conclusions should be promoted into stable AI development guidance.

The file

Show its rules and text

We split this skill into the rules it gives an agent

14 rules. About 0 are common, 9 are distinctive, and 5 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.

  • 5

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this skill to turn user corrections and repeated AI development failures into stable project guidance.

    line 10
  2. Distinctive:

    Use this skill when:

    line 14
  3. Distinctive:

    Check existing rules, project/domain indexes, skills, README, and session-state before creating a new asset.

    line 27
  4. Very specific:

    If the target is .agents/skills/, use .agents/skills/skill-maintainer/SKILL.md.

    line 36
  5. Very specific:

    If the target is .agents/project/ or .agents/domain/, update the corresponding README recall index with owned boundary, read trigger, and …

    line 37
  6. Very specific:

    If the change is still session-local, update .agents/session-state/<session_id>.json instead of stable docs.

    line 38
  7. Distinctive:

    Run the relevant design review and verification gates before final handoff.

    line 39
  8. Distinctive:

    Do not promote a one-off preference into reusable rules.

    line 43
  9. Distinctive:

    Do not put project-specific facts into generic Agent Feed rules.

    line 44
  10. Very specific:

    Do not put reusable AI workflow gates into .agents/project/ or .agents/domain/.

    line 45
  11. Distinctive:

    Do not silently write durable guidance when the user only asked for read-only analysis.

    line 46
  12. Distinctive:

    Do not preserve a user correction as transcript text.

    line 47
  13. Very specific:

    When the correction reveals an external/custom skill gap, route through .agents/skills/specialist-router/SKILL.md or …

    line 48
  14. Distinctive:

    When proposing or applying a promotion, report:

    line 52

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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

Guidance Promoter

Use this skill to turn user corrections and repeated AI development failures into stable project guidance.

Required Use

Use this skill when:

  1. The user corrects how the AI should work in this repository.
  2. The user says a rule, workflow, or conclusion should be remembered.
  3. A repeated failure appears across multiple turns.
  4. Session-state carry-forwards become stable enough to promote.
  5. The user asks to improve .agents, AGENTS, rules, skills, or development protocol docs.

Workflow

  1. Identify what the user corrected, requested to remember, or showed through repeated failure.
  2. State the AI failure, missing source-of-truth, unclear trigger, or workflow gap it reveals.
  3. Decide whether the conclusion is durable guidance or only session-local state.
  4. Check existing rules, project/domain indexes, skills, README, and session-state before creating a new asset.
  5. Choose the owning asset:
    • .agents/rules/: reusable invariant, gate, or protocol constraint.
    • .agents/project/: repository-specific engineering constraint, source layout, verification, release, security, dependency, or delivery boundary.
    • .agents/domain/: stable project concepts, contracts, business rules, source-of-truth ownership, or durable domain facts.
    • .agents/skills/: repeatable task method with clear triggers and workflow steps.
    • .agents/session-state/<session_id>.json: active handoff fact that is useful now but not stable enough to promote.
    • README.md or docs/: user-facing capability, setup, command, or protocol explanation.
  6. Propose the target asset and exact change. Ask for confirmation when .agents/rules/decision-gates.md says the change affects future development results, source-of-truth ownership, public behavior, protocol rules, verification, release, or product scope.
  7. If the target is .agents/skills/, use .agents/skills/skill-maintainer/SKILL.md.
  8. If the target is .agents/project/ or .agents/domain/, update the corresponding README recall index with owned boundary, read trigger, and evidence expectation.
  9. If the change is still session-local, update .agents/session-state/<session_id>.json instead of stable docs.
  10. Run the relevant design review and verification gates before final handoff.

Guardrails

  1. Do not promote a one-off preference into reusable rules.
  2. Do not put project-specific facts into generic Agent Feed rules.
  3. Do not put reusable AI workflow gates into .agents/project/ or .agents/domain/.
  4. Do not silently write durable guidance when the user only asked for read-only analysis.
  5. Do not preserve a user correction as transcript text; extract the repeatable rule, trigger, or source-of-truth boundary.
  6. When the correction reveals an external/custom skill gap, route through .agents/skills/specialist-router/SKILL.md or .agents/skills/skill-maintainer/SKILL.md instead of making the external skill authoritative.

Output

When proposing or applying a promotion, report:

  1. User feedback or repeated failure.
  2. Stable conclusion.
  3. Target asset.
  4. Why that asset owns it.
  5. Index or mirror updates required.
  6. Verification command.

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.