agentleFS
chuckplayerchuckplayer/claude-agent-pack/skills/conventions/SKILL.md

conventions

Skill8 starsChanged 35 days agoMITView on GitHub

Scaffold or update docs/CONVENTIONS.md by reading the actual codebase and interviewing the user. Use when setting up a new project or when conventions have drifted and need to be documented. Trigger this when someone says: document our team standards, set up conventions, what are our coding standards, conventions are out of date, update our style guide, write our CONVENTIONS.md. Do NOT use to read or apply existing conventions — those are loaded automatically. Do NOT use for architectural decisions — use /memory-audit instead.

18 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use review and release.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • SentrySkill

    review

    Fixture review skill.

    Also in 3 projects that use this

  • MicrosoftSkill

    release

    Use ONLY when the user explicitly asks to cut/publish a Huabu desktop release or tag a version. Performs an outward action — bumps the version in apps/desktop/package.json, commits + annotated-tags vX.Y.Z, and pushes the branch + tag. That tag push auto-builds the macOS/Windows installers and publishes a GitHub Release. Provides `status`, `bump`, `tag-push`, and a one-command `run` (bump → tag-push).

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

9 rules. 0 are common, 5 are distinctive, and 4 are very specific to this project.

  • 0

    Common

    Most of the 15 edited versions we read keep these.

  • 5

    Distinctive

    This author's own choices. Worth a look.

  • 4

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Yes: read it fully. This is an update run.

    line 15
  2. Very specific:

    Check memory//*.md for any architecture or design decisions that should be reflected in conventions.

    line 18
  3. Distinctive:

    Use Glob and Grep to observe what the code actually does — conventions should describe reality, not aspiration.

    line 22
  4. Distinctive:

    After gathering all answers:

    line 44
  5. Very specific:

    If creating fresh: write the full file using docs/CONVENTIONS.template.md as the structure, filling every section with what you discovered …

    line 46
  6. Distinctive:

    If updating: edit only the sections that changed.

    line 47
  7. Very specific:

    If the conventions session revealed significant architectural decisions that aren't already in memory/decisions/, note them and ask the …

    line 61
  8. Very specific:

    No template found: If docs/CONVENTIONS.template.md doesn't exist during a fresh scaffold, fall back to the structure used in step 2 …

    line 68
  9. Distinctive:

    Conventions file already exists but is empty: Treat an empty file the same as a missing template — run a fresh scaffold, not an update.

    line 69

“Common” is measured: kept by at least 6 in 10 of the 15 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

Conventions

Scaffold or update docs/CONVENTIONS.md by grounding it in the actual codebase — not templates or guesses.

This skill is interactive. It reads existing code to discover what conventions are already in use, then asks the user to confirm, correct, or extend them.

1. Check current state

  • Does docs/CONVENTIONS.md already exist?
    • Yes: read it fully. This is an update run. Identify which sections need revision.
    • No: this is a fresh scaffold. Read docs/CONVENTIONS.template.md if it exists for the expected structure.
  • Check memory/**/*.md for any architecture or design decisions that should be reflected in conventions.

2. Read the codebase to discover actual conventions

Use Glob and Grep to observe what the code actually does — conventions should describe reality, not aspiration. Sample across all layers:

CategoryWhat to look for
NamingClass/method/variable casing, file naming, interface prefix (I?)
ArchitectureDirectory layers, dependency flow (DI? static?), repo/service/controller separation
Error handlingCatch/log patterns, global error handler, return type (exception/Result<T>/tuple)
TestingFramework, file location pattern, test class naming, mocks vs. real deps
FrontendOptions vs. Composition API, state management, API call pattern (fetch/axios/generated)
OtherLogging levels, auth patterns, compliance requirements visible in code

3. Interview the user

For each category where you found ambiguity, inconsistency, or nothing at all, ask the user one focused question at a time. Do not ask all questions at once — work through them conversationally.

Examples (each asked separately, not together):

  • "I see interface files in the codebase — should they be named IFooService or FooService?"
  • "What's your preferred approach for error handling — exceptions, Result types, or something else?"
  • "Should tests use mocks, real dependencies, or does it depend on the test type?"

4. Write or update CONVENTIONS.md

After gathering all answers:

  • If creating fresh: write the full file using docs/CONVENTIONS.template.md as the structure, filling every section with what you discovered and what the user confirmed.
  • If updating: edit only the sections that changed. Do not rewrite sections that are already accurate. Preserve any user-written content verbatim unless the user asked to change it.

The file must describe actual conventions — not aspirational ones. If a convention is aspirational ("we plan to migrate to X"), mark it clearly:

> **Aspirational:** This convention is the target state. Existing code may not yet comply.

5. Confirm with the user

Present a summary of what was written or changed. Ask: "Does this accurately reflect how you want the team to work?" Make any corrections requested before finishing.

6. Suggest memory updates

If the conventions session revealed significant architectural decisions that aren't already in memory/decisions/, note them and ask the user whether to create memory entries for them. Do not write memory files directly — this skill is read/write only for docs/CONVENTIONS.md.

Gotchas

  • Documenting aspirations as facts: The most common failure mode is writing conventions that describe how the team wants to work rather than how it actually works. If the codebase contradicts a stated convention, flag the inconsistency — do not silently adopt the aspirational version. Mark aspirational items explicitly with the blockquote template in step 4.
  • Asking all questions at once: Step 3 says one question at a time. Dumping a list of questions causes the user to give shallow answers. Work through them conversationally.
  • Rewriting user-written content: On an update run, only touch sections that are incorrect or incomplete. Preserve any user-written prose verbatim unless explicitly asked to change it.
  • No template found: If docs/CONVENTIONS.template.md doesn't exist during a fresh scaffold, fall back to the structure used in step 2 (Naming, Architecture, Error handling, Testing, Frontend, Other) rather than inventing a format.
  • Conventions file already exists but is empty: Treat an empty file the same as a missing template — run a fresh scaffold, not an update.

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.