adr-kit
CLAUDE.md13 starsChanged 4 months agoMITView on GitHub
Read .adr-kit/ADR-guide.md before architectural changes. Architecture decisions live in docs/adr/. Use /adr-kit:context before implementation, /adr-kit:adr for new decisions, and /adr-kit:judge before commit. <!-- ADR-KIT CLAUDE END --> This repository uses Backlog.md (backlog/) as the source of truth for work. - Search for an existing task before starting any meaningful implementation or design work. - If no task exists, create one before editing code or docs. - Prefer the Backlog MCP tools for reading, searching, creating, and updating tasks. Use the…
In this repository19 other files it gives its agents.
The file
We split this CLAUDE.md into the rules it gives an agent
10 rules. About 0 are common, 4 are distinctive, and 6 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
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
- Very specific: line 4
Read .adr-kit/ADR-guide.md before architectural changes.
- Distinctive: line 11
Search for an existing task before starting any meaningful implementation or design work.
- Distinctive: line 13
If no task exists, create one before editing code or docs.
- Very specific: line 14
Prefer the Backlog MCP tools for reading, searching, creating, and updating tasks.
- Distinctive: line 16
Keep tasks small enough to complete in one focused pass.
- Very specific: line 17
Do not edit files in backlog/tasks/ directly.
- Distinctive: line 23
Never switch branches during a test run.
- Very specific: line 35
If graphify-out/wiki/index.md exists, use it for broad navigation instead of raw source browsing.
- Very specific: line 36
Read graphify-out/GRAPH_REPORT.md only for broad architecture review or when query/path/explain do not surface enough context.
- Very specific: line 37
After modifying code, run graphify update .
≈ “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
<!-- ADR-KIT CLAUDE START -->ADR Kit
Read .adr-kit/ADR-guide.md before architectural changes. Architecture decisions live in docs/adr/. Use /adr-kit:context before implementation, /adr-kit:adr for new decisions, and /adr-kit:judge before commit.
<!-- ADR-KIT CLAUDE END -->Backlog.md
This repository uses Backlog.md (backlog/) as the source of truth for work.
- Search for an existing task before starting any meaningful implementation or design work.
- If no task exists, create one before editing code or docs.
- Prefer the Backlog MCP tools for reading, searching, creating, and updating
tasks. Use the
backlogCLI only as a fallback. - Keep tasks small enough to complete in one focused pass.
- Do not edit files in
backlog/tasks/directly.
Working in this repo
- Reproduce the cause a backlog task names before you design a fix; correct the record afterwards. In two sweeps the recorded cause held for 3 of 4 and 4 of 12 tasks respectively. Also check the severity estimate and the completeness of any enumeration, not only the stated cause.
python -m pytest -qtakes ~11 minutes and is mandatory before a merge or release. Targeted runs miss guards in files your change does not touch: a module split broke a check intests/test_setup_project_command.py.- Never switch branches during a test run. A run that starts on branch A and ends on B reports failures that are artifacts of the checkout.
- Make all changes in canonical files first, then regenerate once with
python scripts/build-client-adapters.py, and confirm with--check(expectchanged=0). Never editcodex/orcopilot/by hand. python bin/adr relate ADR-A --to ADR-Bwrites both sides; ADR-028's gate does not let a one-sided link pass lint.python bin/adr accept ADR-Nrefuses without--confirm: acceptance puts a name into an immutable history and must be explicitly requested.- ADR-010 caps lines per file, enforced in
tests/test_release_allowlist.py: entrypoints inscripts/at most 300 lines, support modules 400. Split along the seam the docstring already describes and add the new module to that list. mainanddevare protected withenforce_admins: true, sogh pr merge --admindoes not bypass the requiredvalidatecheck. Use--auto: it weakens nothing and merges once the checks pass. Merging tomainis the maintainer's action.
graphify
This project has a knowledge graph at graphify-out/ with god nodes, community structure, and cross-file relationships.
Rules:
- For codebase questions, first run
graphify query "<question>"when graphify-out/graph.json exists. Usegraphify path "<A>" "<B>"for relationships andgraphify explain "<concept>"for focused concepts. These return a scoped subgraph, usually much smaller than GRAPH_REPORT.md or raw grep output. - If graphify-out/wiki/index.md exists, use it for broad navigation instead of raw source browsing.
- Read graphify-out/GRAPH_REPORT.md only for broad architecture review or when query/path/explain do not surface enough context.
- After modifying code, run
graphify update .to keep the graph current (AST-only, no API cost).
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.

