agentleFS
Sign inStart free
shakhruzshakhruz/mila-core/CLAUDE.md

mila-core

CLAUDE.md3 starsChanged 12 days agoApache-2.0View on GitHub

You are in Mila Agent — the open-source business agent for Telegram. It is a kit for Claude Code: skills (skills/), hooks and tools (install/, templates/home/), working principles (canon/) and the values every Mila reads first (CONSTITUTION.md). The Telegram channel is a separate plugin, mila-telegram. 1. Check what they have: macOS or Linux (Windows only inside WSL), claude --version, bun --version, git. 2. Install: ./install/install.sh --dry-run first and show them what it will touch, then ./install/install.sh. A fresh Linux server:…

  • Installs packages

The file

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

6 rules. About 0 are common, 1 is distinctive, and 5 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.

  • 5

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  • could be automatic: a setting could enforce it
  1. Very specific:

    Never put a real token, key, phone, id, name or client data into a file, a test, a commit or an example.

    line 21
  2. Very specific:

    Never weaken the send-guard (templates/home/.claude/hooks/send-guard.py) or the owner's-yes rule: nothing goes out on the owner's behalf …

    line 23
  3. Very specific:

    Never rewrite CHANGELOG.md history or anything under legacy/ (old names kept for updates).

    line 26
  4. Distinctive:

    Never push, tag or publish — a maintainer does that after review.

    line 27
  5. Very specific:

    Run the tests three ways before you call anything done: bash tests/run_all.sh, CI=true bash tests/run_all.sh, bash tests/run_ci_local.sh …

    line 31
  6. Very specific:

    Every SKILL.md starts with YAML: name and an English description («Use when…», a Russian line may follow).

    line 36

≈ “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 — for Claude Code working in this repository

You are in Mila Agent — the open-source business agent for Telegram. It is a kit for Claude Code: skills (skills/), hooks and tools (install/, templates/home/), working principles (canon/) and the values every Mila reads first (CONSTITUTION.md). The Telegram channel is a separate plugin, mila-telegram.

If a person asked you to install Mila for them

  1. Check what they have: macOS or Linux (Windows only inside WSL), claude --version, bun --version, git.
  2. Install: ./install/install.sh --dry-run first and show them what it will touch, then ./install/install.sh. A fresh Linux server: see deploy/README.md.
  3. Ask the person — and only them — for:
    • a Telegram bot token from @BotFather (they paste it into mila onboard, which hides it);
    • their own Claude sign-in: they run claude and sign in on Anthropic's page themselves, or give their own Anthropic API key. You never ask for, copy, log or store a Claude login or token;
    • pairing: mila onboard shows a 6-character code; they send it to their bot from Telegram.
  4. Start: mila. If the bot is silent: mila doctor.

What you never do here

  • Never put a real token, key, phone, id, name or client data into a file, a test, a commit or an example. Tests use the fake values from tests/fixtures/fake_people.py; token-shaped strings are glued at run time.
  • Never weaken the send-guard (templates/home/.claude/hooks/send-guard.py) or the owner's-yes rule: nothing goes out on the owner's behalf without the owner's explicit yes for that step.
  • Edit CONSTITUTION.md only on the owner's word, with a line in CHANGELOG — its text is approved by the owner and must stay byte-equal across all Milas.
  • Never rewrite CHANGELOG.md history or anything under legacy/ (old names kept for updates).
  • Never push, tag or publish — a maintainer does that after review.

Changing the code

  • Run the tests three ways before you call anything done: bash tests/run_all.sh, CI=true bash tests/run_all.sh, bash tests/run_ci_local.sh (clean HOME, UTC, the workflow's Python packages).
  • A new test that needs a library → add it to the pip install line in .github/workflows/tests.yml the same day.
  • python3 tools/privacy_lint.py --all (no live values) and python3 tools/kitchen_lint.py (no internal task codes or message numbers in public files) must both be clean.
  • Every SKILL.md starts with YAML: name and an English description («Use when…», a Russian line may follow); python3 tools/check_frontmatter.py skills checks that the header parses.
  • Small, reviewable changes; one idea per commit; say in the commit what a person gets from it.

See CONTRIBUTING.md and SECURITY.md.

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.

Check your setup, free