agentleFS
lucasbaburlucasbabur/cutokyo/AGENTS.md

cutokyo

AGENTS.md19 starsChanged 39 days agoApache-2.0View on GitHub

Cutokyo is a local-only observability and control surface for Claude Code, Codex and OpenCode: one Rust application with a CLI and a Tauri desktop. Read README.md for the product and docs/ for depth. - Load the contract before touching domain, ingest, adapters, MCP, config or persistence: .claude/skills/cutokyo-contract/SKILL.md. Its rules (single writer, spool-only hooks, provenance, recovery before every native-config write, owned-entries-only removal) are not negotiable. - Never read or copy the predecessor at ../cutokyo; see docs/provenance/clean-room-policy.md. - cutokyo-domain does no I/O.…

  • Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.

    Of the AGENTS.md files we read, the 33,315 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.

  • In this repository4 other files it gives its agents.

The file

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

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

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Never read or copy the predecessor at ../cutokyo.

    line 13
  2. Distinctive:

    Run what your change touches.

    line 32
  3. Distinctive:

    Keep UI text minimal: no eyebrows, disclaimers or explanatory badges.

    line 66

≈ “Common” is an estimate: at least 1 in 5 of the 42,283 AGENTS.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
AGENTS.md

Agent guide

Cutokyo is a local-only observability and control surface for Claude Code, Codex and OpenCode: one Rust application with a CLI and a Tauri desktop. Read README.md for the product and docs/ for depth.

Before changing code

  • Load the contract before touching domain, ingest, adapters, MCP, config or persistence: .claude/skills/cutokyo-contract/SKILL.md. Its rules (single writer, spool-only hooks, provenance, recovery before every native-config write, owned-entries-only removal) are not negotiable.
  • Never read or copy the predecessor at ../cutokyo; see docs/provenance/clean-room-policy.md.
  • cutokyo-domain does no I/O. Frontends call cutokyo-core::app use cases and never touch the store directly.

Layout

PathWhat lives there
crates/cutokyo-domainPure values, rules, provenance, ports
crates/cutokyo-coreapp, ingest, store (SQLite + FTS5), adapters, mcp, inventory_management
crates/cutokyo-cliThe cutokyo command
crates/cutokyo-desktopTauri shell; service.rs maps use cases to commands
ui/React desktop UI; src/fixtures drive it in a browser without Tauri
tests/Cross-crate pipeline, upgrade and end-to-end tests
tools/fleet/cutokyo-gates.pyStatic contract gates

Checks

Run what your change touches; all of it before a commit.

python3 tools/fleet/cutokyo-gates.py selftest && python3 tools/fleet/cutokyo-gates.py all --root .
cargo fmt --all -- --check
cargo clippy --workspace --all-targets --locked -- -D warnings
cargo test --workspace --locked
pnpm --dir ui check            # types, lint, format, unit tests, knip
pnpm --dir ui test:e2e:web     # Playwright on fixtures; needs port 4173 free
  • Three tests/e2e.rs release tests (release_artifact_manifest, npm_wrapper_install, release_contract_mutations) need a committed tree and fail on uncommitted work. That is expected.
  • pnpm --dir ui test:e2e:tauri builds the real app and drives real windows on the host's Hyprland session. Ask before running it on someone's desktop. It builds with the workspace's pinned @tauri-apps/cli (pnpm exec tauri).
  • On macOS, run cargo test with TMPDIR="$(cd "$TMPDIR" && pwd -P)". /var is a symlink to /private/var, and tests hand temp paths straight to code that refuses symlinked ancestors; production resolves its trusted roots first.
  • Commits need git commit -s and the trailer Co-Authored-By: Claude Code <noreply@anthropic.com> (tools/scripts/check-commit-message.py).

UI work

  • Browse any screen with fixtures: VITE_CUTOKYO_BROWSER_FIXTURES=1 pnpm --dir ui dev --host 127.0.0.1 --port 4173, then /?jev_case=<case>#/<route> (cases in ui/src/fixtures/browserAdapter.ts; large-inventory has ~230 tools, real-world scale).
  • The desktop is WebKitGTK, not Chromium. Button padding, focus rings and some layout differ. Check pixel-level details in WebKitGTK (an offscreen WebKit2.WebView from Python gi is enough), not only in Playwright.
  • Font sizes must use the --font-size-* tokens (ui/src/designTokens.test.ts).
  • Keep UI text minimal: no eyebrows, disclaimers or explanatory badges.

Data scale

Real profiles hold ~1 GB of history, ~500k evidence rows and sessions with 16,000+ observations. Never return unbounded evidence lists to the UI or MCP; results carry the first LISTED_EVIDENCE IDs and an observation_count.

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.