agentleFS
farhan0167farhan0167/cv-claw/CLAUDE.md

cv-claw

CLAUDE.md4 starsChanged 6 months agoMITView on GitHub

Guidance for Claude Code working in this repo. cv-claw is a Python CLI that renders resume JSON into a standalone HTML artifact via Jinja2 templates. The primary user is an LLM that runs shell commands: produce JSON, run cv-claw render, hand the user a finished file. No browser, no dev server, no frontend toolchain. Source layout: src/cvclaw/{schema,templates_loader,render,cli}.py plus src/cvclaw/templates/ (bundled templates shipped in the wheel). Read the modules directly when you need details — they have docstrings. You MUST run…

The file

We split this CLAUDE.md into the rules it gives an agent

3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read the modules directly when you need details — they have docstrings.

    line 14
  2. Very specific:

    If make lint still reports issues, fix them by hand and re-run until clean.

    line 28
  3. Very specific:

    After editing under skills/, run make sync-skills (rsync without --delete) and include the .claude/skills/ changes in the same commit.

    line 45

≈ “Common” is an estimate: at least 1 in 5 of the 26,980 CLAUDE.md 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
CLAUDE.md

CLAUDE.md

Guidance for Claude Code working in this repo.

Project shape

cv-claw is a Python CLI that renders resume JSON into a standalone HTML artifact via Jinja2 templates. The primary user is an LLM that runs shell commands: produce JSON, run cv-claw render, hand the user a finished file. No browser, no dev server, no frontend toolchain.

Source layout: src/cvclaw/{schema,templates_loader,render,cli}.py plus src/cvclaw/templates/ (bundled templates shipped in the wheel). Read the modules directly when you need details — they have docstrings.

Dev workflow — MANDATORY after every code change

You MUST run these three commands, in order, after ANY edit to a .py file:

make lint-fix
make format
make lint

If make lint still reports issues, fix them by hand and re-run until clean. Do not commit, hand work back, or report a task complete while make lint is failing. This repo has no test suite — ruff is the only automated quality gate.

Manual verification: uv run cv-claw render resumes/example.json and inspect the output.

Skills live in two places — MANDATORY sync rule

  • skills/ — source of truth. Public; ships to end users. Always edit here.
  • .claude/skills/ — local mirror Claude Code reads while developing this repo. May also contain repo-local-only skills (e.g. pr-message) that have no skills/ counterpart — don't delete those.

After editing under skills/, run make sync-skills (rsync without --delete) and include the .claude/skills/ changes in the same commit. File deletions don't propagate — remove them by hand from both mirrors.

Commit conventions

  • Conventional Commits: <type>(<scope>): <subject>, imperative lowercase, ≤72 chars, no period. Draw scopes from git log -20 rather than inventing new ones.
  • Include a body when the subject can't carry the why on its own. Wrap ~72; lead with motivation.
  • Last line of the body: Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
  • No 🤖 Generated with Claude Code lines on commits (those are for PR bodies). No --amend or --no-verify without an explicit user request.

Schema is load-bearing

Section kind is a fixed taxonomy: prose, keyvalue, list, timeline. Adding a kind is a schema change that ripples through every template. Don't add kinds; map novel layouts onto the existing four with custom CSS or sparse section.name branches.

The authoritative schema is src/cvclaw/schema.py. The skill's schema reference is the single human-readable mirror; update it when the pydantic models change. No other fan-out.

Templates

  • Two roots: workspace ./.cvclaw/templates/ (user-authored) and the bundled src/cvclaw/templates/ (ships in the wheel). Workspace shadows bundled per directory.
  • Per-template CSS rules must start with .<name> to prevent cross-template leakage when CSS is inlined.
  • Directory names beginning with _ are reserved for shared partials and skipped by discovery.

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.