agentleFS
s3onghyuns3onghyun/bulchimbeon/AGENTS.md

bulchimbeon

AGENTS.md59 starsChanged 25 days agoMITView on GitHub

bulchimbeon is a macOS menu-bar app (Swift Package Manager, no Xcode project) that keeps the Mac awake while AI coding agents are running, plus a Linux daemon and CLI written in Go (a Go module at the repository root; Windows is a stub for later). Read CONTRIBUTING.md before changing anything; it is the source of truth for code layout, ground rules, adding an agent, sprites and translations. SECURITY.md holds the threat model. Requires macOS 13 or later (Package.swift targets .macOS(.v13))…

  • 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 32,671 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 repositoryOne other file it gives its agents.

The file

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

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

  • 6

    Very specific

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

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

    Put logic, especially agent detection and duty decisions, in Sources/BulchimbeonCore/ with tests in Tests/BulchimbeonCoreTests/, mirrored …

    line 51
  2. Very specific:

    No network access apart from the opt-out once-a-day update check (UpdateCheck in Core).

    line 54
  3. Very specific:

    Write every user-facing string in both Korean and English: L10n.t("한국어", "English").

    line 56
  4. Very specific:

    After changing Sources/BulchimbeonCore/Sprites.swift, run make icons and commit the regenerated Resources/AppIcon.icns and …

    line 57
  5. Distinctive:

    Keep a PR small and focused.

    line 60
  6. Very specific:

    Treat these as security-sensitive and do not change them incidentally: .github/workflows/, scripts/ (app build, signing, the Homebrew tap …

    line 64
  7. Very specific:

    Never put outside strings into the admin (pmset) command, and never store or print other processes' command lines or environment variables.

    line 69

≈ “Common” is an estimate: at least 1 in 5 of the 41,581 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

AGENTS.md

Scope

bulchimbeon is a macOS menu-bar app (Swift Package Manager, no Xcode project) that keeps the Mac awake while AI coding agents are running, plus a Linux daemon and CLI written in Go (a Go module at the repository root; Windows is a stub for later). Read CONTRIBUTING.md before changing anything; it is the source of truth for code layout, ground rules, adding an agent, sprites and translations. SECURITY.md holds the threat model.

Setup and verification

Requires macOS 13 or later (Package.swift targets .macOS(.v13)) and the Xcode Command Line Tools with a Swift 6 or newer toolchain, because the tests use Swift Testing. Full Xcode is optional. There are no third-party dependencies to install. The app links IOKit and ServiceManagement, so the Swift package does not build on Linux; the Go module below does.

make build   # swift build
make test    # swift test, plus the flags the Command Line Tools need
make app     # ./scripts/build-app.sh -> dist/Bulchimbeon.app (ad-hoc signed)
make icons   # after editing sprites: regenerates Resources/AppIcon.icns and docs/sprites-preview.png

CI (.github/workflows/ci.yml, check name test) runs swift test, builds the app bundle and verifies its signature on macOS. Run make test and make app before opening a pull request. Do not claim a command passed if it was not run; if you are not on macOS or a prerequisite is missing, say so and include any failure output.

Go module (Linux daemon)

Go 1.27 or newer, standard library only (no require in go.mod).

go vet ./... && go test ./...       # what the CI job `go (linux)` runs, plus gofmt
go build ./cmd/bulchimbeon          # run | status | doctor | config
./scripts/package-linux.sh <version> <amd64|arm64>   # tar.gz and .deb into dist/linux
  • Engine truth lives in the Swift core (Sources/BulchimbeonCore). internal/engine is its Go port, and both must pass the shared fixtures in testdata/engine/ (format in its README).
  • To change engine behavior: change the Swift core, update the fixture until swift test passes, then make go test ./... pass. Never edit a fixture just to make the Go port pass.
  • internal/platform/linux holds every Linux system call (/proc, systemd-inhibit, busctl, sysfs); keep internal/daemon and cmd/bulchimbeon free of duty logic. Run external tools by absolute path with a minimal environment, never through a shell, and never print process arguments.
  • The polkit rule is only ever printed (bulchimbeon doctor, README); never run it.

Project conventions

  • Put logic, especially agent detection and duty decisions, in Sources/BulchimbeonCore/ with tests in Tests/BulchimbeonCoreTests/, mirrored in internal/engine through the fixtures. Keep Sources/Bulchimbeon/ a thin shell that calls the system.
  • No network access apart from the opt-out once-a-day update check (UpdateCheck in Core); no telemetry. No third-party dependencies.
  • Write every user-facing string in both Korean and English: L10n.t("한국어", "English").
  • After changing Sources/BulchimbeonCore/Sprites.swift, run make icons and commit the regenerated Resources/AppIcon.icns and docs/sprites-preview.png.
  • Settings keys, launch flags and the duty-log format follow docs/compatibility.md.
  • Keep a PR small and focused. Do not reformat unrelated files.

Security-sensitive paths

Treat these as security-sensitive and do not change them incidentally: .github/workflows/, scripts/ (app build, signing, the Homebrew tap bump, Linux packaging and install), Package.swift, go.mod, Makefile, SECURITY.md, and anything touching releases or the Homebrew tap. Never add secrets, tokens, a new third-party Action, or a workflow trigger such as pull_request_target. Never put outside strings into the admin (pmset) command, and never store or print other processes' command lines or environment variables.

AI-assisted work

AI assistance is welcome for focused, reviewed work. The human submitting the pull request is responsible for every line, test result, license/copyright implication, and follow-up. Disclose material AI assistance in the PR template; do not submit unreviewed bulk-generated changes or claim tests you did not run.

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.