dtg-claude-plugins
CLAUDE.md0 starsChanged 7 months agoMITView on GitHub
Monorepo — see each packages/*/CLAUDE.md for package-specific details. Releases are automated via release-please. Use the release-flow skill to drive the full dev→main flow; the Release PR is always manually merged after review. Merge-time checks referenced by /merge-pr (Design Doc Deviation Check, Pre-merge Checks) are defined in docs/merge-checks.md. After modifying source code (src/ or tests/), always run the following before committing: Do not commit until both commands pass cleanly. When creating issues with gh issue create, always attach the appropriate labels:
In this repositoryOne other file it gives its agents.
- release-flowSkill
The file
We split this CLAUDE.md into the rules it gives an agent
7 rules. About 0 are common, 2 are distinctive, and 5 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.
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
- Distinctive: line 5
All project assets MUST be written in English:
- Very specific: line 25
Never push directly to main or dev.
- Very specific: line 31
Commit messages MUST follow Conventional Commits.
- Very specific: line 41
After modifying source code (src/ or tests/), always run the following before committing:
- Distinctive: line 46
Do not commit until both commands pass cleanly.
- Very specific: line 50
When creating issues with gh issue create, always attach the appropriate labels:
- Very specific: line 57
When creating PRs or issues, check .github/ for templates and follow their format if present.
≈ “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.
Files
CLAUDE.md - dtg-claude-plugins
Language Policy
- All project assets MUST be written in English:
- CLAUDE.md
- Skills (SKILL.md)
- Agents (.md)
- Knowledge (.md)
- Source code comments and log messages
- Schemas (descriptions, titles)
- Plugin manifests (plugin.json)
- User-facing conversation language follows the global CLAUDE.md settings.
Architecture
Monorepo — see each packages/*/CLAUDE.md for package-specific details.
Branch Strategy
main: Production branch. Plugin users consume this branch directly.dev: Development branch. All feature work starts here.- Feature branches are created from
devand merged back intodevvia PR. devis merged intomainvia PR when ready to release.- Never push directly to
mainordev.
Release Flow
Releases are automated via release-please. Use the release-flow skill to drive the full dev→main flow; the Release PR is always manually merged after review.
- Commit messages MUST follow Conventional Commits. Scope must match a package component (
runnel,silo, …):feat(<scope>):→ minor,fix(<scope>):→ patch,!orBREAKING CHANGE:→ major (minor while0.x). - Local
npx tsupis not required —dist/is rebuilt by CI at release time.
Merge Checks
<!-- Do not rename this heading — the /merge-pr skill uses it as a search key. -->Merge-time checks referenced by /merge-pr (Design Doc Deviation Check, Pre-merge Checks) are defined in docs/merge-checks.md.
Pre-commit Verification
After modifying source code (src/ or tests/), always run the following before committing:
pnpm lint:fix— auto-fix lint issues across all packagespnpm check— runs typecheck, lint, oxlint, and tests for all packages
Do not commit until both commands pass cleanly.
Issue Labeling
When creating issues with gh issue create, always attach the appropriate labels:
- Plugin label (required):
plugin:runnel. Omit only for repo-wide issues. - Type label:
bugfor defects,enhancementfor new features or improvements.
GitHub Templates
When creating PRs or issues, check .github/ for templates and follow their format if present.
Reference
- Claude CLI verified behavior — empirically verified
claude -pflags, skill discovery, invocation triggers, etc.
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
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.

