Index of every page on this site: llms.txt
All skills
dominiktobureto avatar

/repo-health-check

@9125e02
Run it yourself

Use when auditing a repo before changes to find the smallest safe PR, quality risks, stale docs, missing tests, ignored-file gaps, or agent setup issues. Best first skill for unfamiliar codebases.

  • 2 files
  • 5.5 KB
  • Updated 6 months ago
  • GitHub
Use this Skill: https://skilld.dev/gh/dominiktobureto/awesome-grok-build/repo-health-check

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

SKILL.md

≈54 tokens for metadata: the name and description. ≈642 when used: this file. ≈699 more on demand in 1 file.

Description uses 2.7% of example budget

Before choosing a Skill, your Agent reads its name and description. All available Skills share that space.

  • A shorter description leaves more room for other Skills. This entry exceeds our 1% size suggestion.

In our Claude Code example, all Skill names and descriptions share 8,000 characters. This Skill uses ≈216 characters, or 2.7%.

The 1% threshold is a size suggestion. Longer descriptions can still fit.

Your model, settings, and other Skills decide how much text your Agent can read.

Example settings and source

The example uses a 200k-token context and default Claude Code settings. The count includes the name, description, separators, and when_to_use when present. Codex also counts local file paths.

Skit's source and limits: Codex 0.160.1, Claude Code 2.1.292.

Repo Health Check

Find the smallest useful change with the best risk/reward ratio.

Grok Build Mode

  • Start in Plan Mode for unfamiliar repos or when changes may touch multiple subsystems.
  • Use subagents for parallel read-only discovery when the repo is medium/large:
    • structure: map stack, entry points, and package managers.
    • quality: find test, lint, typecheck, CI, and coverage signals.
    • docs: compare README commands with real config.
    • safety: inspect .gitignore, .grokignore, secrets patterns, hooks, and AGENTS.md.
  • Arena-style comparison: if Grok Build exposes Arena Mode, run candidate-first-PR proposals against 3-5 agents and rank by verification confidence. Otherwise simulate by asking subagents for independent candidates.

Workflow

  1. Read repo instructions: AGENTS.md, README.md, package/config files, CI files.
  2. Identify project type, main runtime, package manager, and quality gates.
  3. Look for high-signal risks:
    • stale or missing setup docs;
    • absent tests around core behavior;
    • failing lint/type/test commands;
    • generated output committed unexpectedly;
    • missing .grokignore or obvious secret exposure;
    • dependency and lockfile mismatch;
    • unsafe hooks or unclear agent instructions.
  4. Produce 3 candidate PRs ranked by impact, size, blast radius, and verification confidence.
  5. Ask for human approval before edits.
  6. Implement only the chosen candidate.
  7. Run the smallest relevant lint/test/type/docs check.
  8. Summarize changed files and residual risks.

Git Discipline

  • Never commit generated junk, secrets, or local agent state.
  • Keep the first PR tiny.
  • If tests fail for unrelated reasons, capture the exact failure and avoid broad cleanup.

Example Prompts

Use repo-health-check. Inspect this repo and propose the smallest safe PR we should ship first. Do not edit yet.
Run repo-health-check with subagents for structure, tests, docs, and safety. Rank the top 3 fixes and wait for my approval.

Output

Start with:

Bottom line: <one sentence>

Then provide:

  • repo type and stack;
  • discovered commands;
  • top risks;
  • 3 candidate PRs;
  • recommended first PR;
  • verification plan;
  • whether edits require Plan Mode approval.

Guardrails

  • Do not invent test commands. Read package/config files first.
  • Do not edit generated files unless the repo intentionally tracks them.
  • Do not upgrade dependencies unless explicitly requested.
  • If no safe command is available, explain what could not be verified.

Source: SKILL.md on GitHub

No rule matched.

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

No third-party reports yet.

Signed by skilld at 9125e02. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub last month.

Steadyupdated 6 months ago
version
2.0.0
author
awesome-grok-build

README badge

README badge for dominiktobureto/awesome-grok-build/repo-health-check