YO

yogieat-harness-governor

Promotes recurring project observations into permanent configuration or rules.

Install

mkdir -p .claude/skills/yogieat-harness-governor && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/13048" && unzip -o skill.zip -d .claude/skills/yogieat-harness-governor && rm skill.zip

Installs to .claude/skills/yogieat-harness-governor

Activation

This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.

Use when repeated Yogieat prompts, review comments, or misunderstandings should be translated into harness candidates or stable rules. Classifies new observations, applies the candidate-to-promotion policy, and keeps harness updates intentional rather than noisy.
263 chars✓ has a “when” triggerlonger than Claude Code's old 250-char listing cap (fine on current versions)
Intermediate

Key capabilities

  • Classify new observations.
  • Register new candidate rules.
  • Promote candidates to stable rules.
  • Propose harness updates during feature work.
  • Update relevant rule, skill, agent, or governance files.

How it works

The skill reads governance documents, classifies new observations as stable rules, new candidates, or local preferences, and applies a promotion policy to update the harness intentionally.

Inputs & outputs

You give it
A repeated instruction, recurring review finding, or stable preference.
You get back
A classified observation, a new candidate rule, or a promoted stable rule.

When to use yogieat-harness-governor

  • Promote repeat instructions
  • Update project rules
  • Refine validation scope

About this skill

Yogieat Harness Governor

Use this skill when the conversation reveals a repeated instruction, a recurring review finding, or a stable preference that should live in the harness.

Workflow

  1. Read ./AGENTS.md, .codex/rules/governance.md, and .codex/governance/rule-candidates.md.
  2. Classify the new observation as:
    • already covered by a stable rule
    • new candidate rule
    • candidate ready for promotion
    • task-local preference that should stay out of the harness
  3. Prefer candidate registration before promotion unless the miss is high severity.
  4. During unrelated feature work, propose harness updates instead of silently editing them.
  5. During dedicated harness work, update the relevant rule, skill, agent, or governance file directly.

Good Candidates

  • Repeated service or facade boundary clarifications
  • Repeated validation scope corrections
  • Repeated transaction or test harness pitfalls
  • Repeated long prompt boilerplate that a skill or rule can replace

Bad Candidates

  • One-off naming tastes
  • Isolated wording preferences
  • Non-recurring refactor choices

Notes

  • Promotion threshold: two repeats, or one high-risk miss.
  • Stable rules should stay short and operationally meaningful.

When not to use it

  • For one-off naming tastes.
  • For isolated wording preferences.
  • For non-recurring refactor choices.

Limitations

  • It does not handle one-off naming tastes.
  • It does not handle isolated wording preferences.
  • It does not handle non-recurring refactor choices.

How it compares

This skill formalizes the process of incorporating repeated feedback into a stable harness, preventing ad-hoc changes and ensuring intentional updates, unlike manually editing rules based on individual instances.

Compared to similar skills

yogieat-harness-governor side by side with the closest alternatives in the catalog.

SkillInstallsUpdatedSafetyDifficulty
yogieat-harness-governor (this skill)02moNo flagsIntermediate
evolving-skill-creator17moNo flagsBeginner
tsg-pipeline01moNo flagsIntermediate
blueprint-builder01moNo flagsAdvanced

Try saying

Example prompts that trigger this skill in your AI assistant.

Search skills

Search the agent skills registry