agentleFS
Cody-SimsCody-Sims/agent-skills/skills/codebase-exploration/SKILL.md

codebase-exploration

Skill0 starsChanged 4 months agoMITView on GitHub

Maps unfamiliar code and locates change sites, scoped instructions, relevant tests, and validation commands. Use when onboarding, tracing an unknown change site, or explaining repository structure; not for a known-file edit with sufficient context. Stays read-only.

6 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

9 rules. 0 are common, 9 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 6 edited versions we read keep these.

  • 9

    Distinctive

    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 the applicable repository and path-scoped instructions.

    line 20
  2. Distinctive:

    Follow the relevant entry point or symbol to its callers, dependencies, and nearest tests.

    line 23
  3. Distinctive:

    Prefer code intelligence, then scoped file discovery and text search.

    line 26
  4. Distinctive:

    Keep a single continuous trace local.

    line 32
  5. Distinctive:

    Stop when the question is answered or the edit site, dependent call sites, relevant tests, and validation commands are known.

    line 36
  6. Distinctive:

    Use the exploration playbook only when the layout or ecosystem is unclear.

    line 40
  7. Distinctive:

    Do not edit files, change git or remote state, or install dependencies.

    line 46
  8. Distinctive:

    Prefer inspection to execution.

    line 47
  9. Distinctive:

    If answering requires mutation, report that boundary rather than performing it.

    line 51

“Common” is measured: kept by at least 6 in 10 of the 6 edited versions of this file we could read. “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

Codebase Exploration

Answer the bounded question or find the smallest correct edit site. Do not turn local orientation into a full repository audit.

Workflow

  1. Define the question and stop condition. Reuse current context and prior evidence; revisit it only when changed, missing, or contradictory.
  2. Read the applicable repository and path-scoped instructions. Locate declared commands in manifests or task runners. Discover optional instruction/reference paths before reading; do not probe every filesystem ancestor or guess names.
  3. Follow the relevant entry point or symbol to its callers, dependencies, and nearest tests. Learn the local conventions and preservation requirements. Map unrelated boundaries only when the user requests a repository-wide map.
  4. Prefer code intelligence, then scoped file discovery and text search. Batch independent reads; request matching paths before contents, and ranges before full large files. Reuse confirmed absence within the same unchanged scope; retry only after new evidence or a relevant filesystem change. Missing required guidance is a blocker, not permission to ignore it. A denied read is not absence: never retry it through another tool. Widen only for a concrete evidence gap.
  5. Keep a single continuous trace local. Delegate only substantial independent questions, not work answerable in two to five direct calls. Give each worker a question, scope, stop condition, and compact evidence contract. Consume its report instead of repeating the same exploration.
  6. Stop when the question is answered or the edit site, dependent call sites, relevant tests, and validation commands are known. Report any unresolved gap rather than inventing an architecture or continuing an open-ended scan.

Use the exploration playbook only when the layout or ecosystem is unclear. Its repository-wide order is not mandatory for a local question.

Boundaries

  • Do not edit files, change git or remote state, or install dependencies.
  • Prefer inspection to execution. Run a build, test, or lint only if it does not alter tracked files, install packages, or write outside build directories.
  • For a running-system symptom, check relevant service readiness before treating startup lag as a code defect; do not perform an unrelated bootstrap audit.
  • If answering requires mutation, report that boundary rather than performing it.

Output

Return a compact evidence map: scope, repository state, change site, applicable instructions, relevant behavior/tests, and exact validation commands with path:line sources. Include boundaries only as needed. Label inference and unknowns; do not present them as verified. Internal orientation need not become a separate user-facing report or an unrequested file.

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.