agentleFS
Cody-SimsCody-Sims/agent-skills/skills/repository-agent-bootstrap/SKILL.md

repository-agent-bootstrap

Skill0 starsChanged 4 months agoMITView on GitHub

Audits and improves repository-owned AGENTS.md, scoped instructions, custom agents, and skill discovery without replacing conventions. Use when setting up or maintaining repository agent guidance, integrating an agent or skill, or fixing conflicting or overly broad customization; not for routine code edits, global preferences, or standalone skill authoring.

The file

We split this skill into the rules it gives an agent

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

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read other bodies only to resolve an overlap, conflict, or stale reference.

    line 25
  2. Distinctive:

    When guidance requires a skill, use the loading contract to distinguish an advertised native name from an authorized repository source.

    line 28
  3. Distinctive:

    Keep AGENTS.md navigational, repository instructions broadly applicable, and scoped instructions limited to accurate path globs.

    line 41
  4. Distinctive:

    Use self-contained, task-oriented skills for on-demand workflows and relative links for specialized resources instead of duplicating them.

    line 43
  5. Distinctive:

    Add an agent only for a distinct role, isolated context, tool boundary, or genuine handoff.

    line 45
  6. Distinctive:

    Update nearby indexes/discovery documents only as repository convention requires.

    line 48
  7. Distinctive:

    Validate the affected discovery path: Check syntax, frontmatter, names, path globs, relative resources/links, and one clear owner per rule.

    line 50
  8. Distinctive:

    Run available narrow customization or documentation checks.

    line 54
  9. Distinctive:

    Do not expand a targeted repair into an unrequested repository bootstrap.

    line 61

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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

Repository Agent Bootstrap

Improve requested repository customization, not every code-change task. Repository guidance is project-owned and shareable; global/user configuration is personal, cross-repository policy. Never edit global configuration here or copy personal preferences unless they are genuine repository requirements. Report global conflicts or missing capabilities separately.

Workflow

  1. Discover within scope. Identify repository root, status, and requested customization target. Inventory root/nested AGENTS.md, repository/scoped instructions, agents, and skills. Read guidance governing the target's path and precedence, plus relevant contributor rules and validation commands. Read other bodies only to resolve an overlap, conflict, or stale reference; a repository-wide customization audit includes all applicable guidance, not unrelated application code or full skill-content audits. When guidance requires a skill, use the loading contract to distinguish an advertised native name from an authorized repository source. Resolve the route in each intended worker context before delegation; do not assume coordinator discovery carries over.
  2. Map the smallest change. State the current model and precedence; identify gaps, duplicate/conflicting rules, stale links, and overly broad loading. Separate project from personal ownership. Propose keep/update/create actions, assigning one owner per rule. Use artifact selection when ownership/type is unclear. Stop for user confirmation of ambiguous ownership or deletion, renaming, or reorganization; setup is not consent to destructive changes.
  3. Implement the approved scope. Improve existing artifacts first; preserve unrelated content, names, locations, terminology, formatting, and precedence.
    • Keep AGENTS.md navigational, repository instructions broadly applicable, and scoped instructions limited to accurate path globs.
    • Use self-contained, task-oriented skills for on-demand workflows and relative links for specialized resources instead of duplicating them.
    • Add an agent only for a distinct role, isolated context, tool boundary, or genuine handoff; explain its unique capability, not a redundant persona. Bootstrap does not authorize launching agents or a default review fleet.
    • Update nearby indexes/discovery documents only as repository convention requires. Do not add competing guidance or automatic per-task audits.
  4. Validate the affected discovery path. Check syntax, frontmatter, names, path globs, relative resources/links, and one clear owner per rule. Ensure no new artifact shadows existing guidance. Record verified skill-loading routes or blockers without changing global installation or permissions. Run available narrow customization or documentation checks; do not invent dependencies just to validate. Inspect the final diff for unrelated or global edits.
  5. Stop and hand off. Finish when the affected discovery path is coherent, specialized content loads on demand, and conventions remain intact, or report a blocked decision/check. Return created/modified paths, check commands and results, unresolved conflicts, and separately recommended global actions. Do not expand a targeted repair into an unrequested repository bootstrap.

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.