agentleFS
SamsungSamsung/tizen-agent-skills/AGENTS.md

tizen-agent-skills

AGENTS.md19 starsChanged 21 days agoApache-2.0View on GitHub

This repository is Samsung's open-source collection of Tizen development plugins for AI coding agents. Each top-level directory is one independently versioned plugin (tizen-sdk-skills/, tizen-action-skills/, …). Start with README.md and CONTRIBUTING.md. For detailed authoring and review instructions, read each plugin's own CONTRIBUTING.md (for example tizen-sdk-skills/CONTRIBUTING.md). - common/ is the single source of truth for every plugin — skills, agents, hooks, lib/, scripts/, setup/. The host wrapper directories (claude/, cline/, codex/, gemini/) hold thin adapters only; never duplicate logic into them. -…

  • Installs packages
  • 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 repository3 other files it gives its agents.

The file

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

11 rules. About 0 are common, 2 are distinctive, and 9 are 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.

  • 9

    Very specific

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

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

    Start with README.md and CONTRIBUTING.md.

    line 7
  2. Very specific:

    Do not edit the mirrored copies under ~/.claude, ~/.cline, ~/.codex or ~/.gemini.

    line 20
  3. Very specific:

    After changing a runner-lookup snippet or HOST_DOT_DIRS in common/lib/core/plugin-cache.js, run node scripts/rewrite-runner-snippets.js and …

    line 22
  4. Very specific:

    Keep plugin.json version and package.json version in sync within a plugin.

    line 34
  5. Distinctive:

    Follow the checklist in CONTRIBUTING.md → Adding a new plugin.

    line 42
  6. Very specific:

    Write English first; provide a Korean mirror (README.ko.md, CONTRIBUTING.ko.md, CHANGELOG.ko.md).

    line 50
  7. Distinctive:

    Use lowercase kebab-case for new file and directory names.

    line 53
  8. Very specific:

    Keep llms.txt updated when adding significant new documentation.

    line 54
  9. Very specific:

    Use Conventional Commits with the plugin name as scope, for example fix(tizen-sdk-skills): ..., docs(tizen-action-skills): ....

    line 65
  10. Very specific:

    Keep pull requests focused on one plugin directory.

    line 68
  11. Very specific:

    No build step or test suite yet.

    line 122

≈ “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

Working in tizen-agent-skills

This repository is Samsung's open-source collection of Tizen development plugins for AI coding agents. Each top-level directory is one independently versioned plugin (tizen-sdk-skills/, tizen-action-skills/, …).

Start with README.md and CONTRIBUTING.md. For detailed authoring and review instructions, read each plugin's own CONTRIBUTING.md (for example tizen-sdk-skills/CONTRIBUTING.md).

Essential rules

Plugin structure

  • common/ is the single source of truth for every plugin — skills, agents, hooks, lib/, scripts/, setup/. The host wrapper directories (claude/, cline/, codex/, gemini/) hold thin adapters only; never duplicate logic into them.
  • Do not edit the mirrored copies under ~/.claude, ~/.cline, ~/.codex or ~/.gemini; re-run the plugin's setup script instead.
  • After changing a runner-lookup snippet or HOST_DOT_DIRS in common/lib/core/plugin-cache.js, run node scripts/rewrite-runner-snippets.js and then --check.

Versioning

  • Repository release version — root package.json version, tracked in root CHANGELOG.md. Tags follow tizen-agent-skills-vX.Y.Z.
  • Plugin versions are independent — each plugin's package.json and common/.claude-plugin/plugin.json carry the plugin's own version (for example tizen-sdk-skills is at 1.4.x, tizen-action-skills at 1.0.x). Bump them independently of the release version.
  • Keep plugin.json version and package.json version in sync within a plugin.
  • marketplace.json top-level version and metadata.version are the catalog version. They follow the tizen-sdk-skills plugin version: bump them together with every tizen-sdk-skills version bump.

Adding a new plugin

Follow the checklist in CONTRIBUTING.md → Adding a new plugin. A new plugin must register in marketplace.json, README.md/README.ko.md plugin table (between <!-- plugins:start --> / <!-- plugins:end -->), llms.txt, CONTRIBUTING.md, .github/CODEOWNERS, and release.yml.

Documentation

  • Write English first; provide a Korean mirror (README.ko.md, CONTRIBUTING.ko.md, CHANGELOG.ko.md).
  • Give each hand-written document one H1 as its first content heading.
  • Use lowercase kebab-case for new file and directory names.
  • Keep llms.txt updated when adding significant new documentation.

Security

  • Never commit internal hostnames, private IP addresses, personal home paths, credentials, certificates (other than the documented throwaway test fixture) or captured session logs that contain them. The pull request template has a checklist item for this.

Commits and pull requests

  • Use Conventional Commits with the plugin name as scope, for example fix(tizen-sdk-skills): ..., docs(tizen-action-skills): ....
  • Keep pull requests focused on one plugin directory; CI workflows are scoped per project (.github/workflows/*.yml filter on <project>/**).
  • Open an issue before starting large changes so the approach can be agreed first.

Validating a change

tizen-sdk-skills

# Lint and format (repo-wide)
pnpm install --frozen-lockfile && pnpm run lint && pnpm run format:check

# Unit tests
node common/lib/tests/run-all.js
bash common/hooks/hooks.test.sh

# Runner snippets and SPDX headers in sync
node scripts/rewrite-runner-snippets.js --check
node scripts/add-spdx-headers.js --check

# Rules on the files this change touches (authoring; add --format jsonl when reviewing,
# CI uses --format github to annotate the pull request's Files changed tab)
node scripts/check-pr.js --base origin/main

# Standalone CLI compiles
cd tizen-cli && pnpm install --frozen-lockfile && pnpm exec tsc --noEmit && pnpm run build

# Test-case schema lint (no SDK or device required)
cd tests && npm ci && node runner.mjs --dry-run
npm run lint        # TC schema dry-run + doc-stats + runner helper tests
npm run test:safe   # safe-tier TCs, no SDK or device required

pnpm is used at the plugin root and in tizen-cli/; npm is used in tests/. Test tiers (safe / mutating / device) are defined in tizen-sdk-skills/tests/README.md. Record which tiers you ran in the PR's Testing section.

check-pr.js reports each finding with a rule ID. ERROR blocks the pull request; fix every one. WARN does not block, but each one must be fixed or explained in the pull request description before approval. A run always ends with a summary line such as check-pr: 0 ERROR, 0 WARN (12 files, 0.05s); if nothing is printed, the check did not run, so check the --base revision.

The check cannot judge these, so its silence is not approval:

  • logic duplicated into a host wrapper instead of common/;
  • a spawned process whose exit code is not checked;
  • a .sh change without the matching .ps1 change (or the reverse);
  • personal information inside an added image or screenshot.

tizen-action-skills

No build step or test suite yet. Verify that plugin.json and package.json versions are in sync and that the skill README renders correctly.

Reviewing a pull request

To review someone else's tizen-sdk-skills pull request, follow .claude/skills/tizen-sdk-skills-review/SKILL.md. It runs check-pr.js and the matching safe-tier checks, rechecks earlier review findings, and posts one English review comment after the reviewer confirms the draft. Agents never approve, request changes or merge.

That file is the single copy of the procedure. Codex and Cline load it through thin entrypoints that only point back to it:

  • Codex: .agents/skills/tizen-sdk-skills-review/SKILL.md
  • Cline: .cline/skills/tizen-sdk-skills-review/SKILL.md

Change the procedure in .claude/skills/ only; do not copy it into the entrypoints.

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.