cognivia
CLAUDE.md1 starsChanged 2 months agoMITView on GitHub
- Enter Claude Code plan mode before architectural, multi-file, migration, or destructive changes so the proposed scope can be reviewed before edits. - Prefer project skills for reusable, multi-step workflows instead of adding those procedures to this file. - Keep custom subagents read-only by default and give them one bounded research or review responsibility. - Do not run parallel agents that edit the same working tree. - Use separate worktrees for parallel editing only when the user explicitly authorizes their…
In this repository9 other files it gives its agents.
- AGENTS.mdAGENTS.md
- architecture-auditSkill
- commit-reviewSkill
- cxpSkill
- docs-updateSkill
- safe-refactorSkill
- session-handoffSkill
- task-briefSkill
- cxpSkill
The file
We split this CLAUDE.md into the rules it gives an agent
7 rules. About 0 are common, 5 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈5
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 7
Prefer project skills for reusable, multi-step workflows instead of adding those procedures to this file.
- Distinctive: line 9
Keep custom subagents read-only by default and give them one bounded research or review responsibility.
- Distinctive: line 11
Do not run parallel agents that edit the same working tree.
- Distinctive: line 12
Use separate worktrees for parallel editing only when the user explicitly authorizes their creation and cleanup.
- Distinctive: line 14
Treat tool output as evidence to inspect, not proof by itself.
- Very specific: line 16
Use docs/agents/README.md as the human-facing guide to this tooling.
- Very specific: line 20
Keep symlink adapters for exactly the six canonical responsibility skills.
≈ “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
@AGENTS.md
Claude Code adapter
- Enter Claude Code plan mode before architectural, multi-file, migration, or destructive changes so the proposed scope can be reviewed before edits.
- Prefer project skills for reusable, multi-step workflows instead of adding those procedures to this file.
- Keep custom subagents read-only by default and give them one bounded research or review responsibility.
- Do not run parallel agents that edit the same working tree.
- Use separate worktrees for parallel editing only when the user explicitly authorizes their creation and cleanup.
- Treat tool output as evidence to inspect, not proof by itself; verify the relevant files, diffs, and exit results.
- Use
docs/agents/README.mdas the human-facing guide to this tooling. - Canonical project skill bodies live under
.agents/skills/. Project entries under.claude/skills/are discovery adapters only; keep canonical adapters as symlinks instead of copyingSKILL.mdbodies. - Keep symlink adapters for exactly the six canonical responsibility skills.
The project-owned
cxpworkflow uses one additional import-only, explicit-only adapter; retired Capstone aliases must not be reintroduced.
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.

