canon-skills
CLAUDE.md9 starsChanged 21 days agoMITView on GitHub
Canon-specific guidance that applies only in this repo. Global instructions are in ~/.claude/CLAUDE.md. When running multiple Claude Code instances in parallel worktrees: - Each worktree uses a long-lived tracking branch (e.g., claude-code-1/main-1) - After a PR merges, reset the tracking branch from inside the worktree: tkt worktree-reset (fetches first, refuses on uncommitted, untracked or unpushed work, shows what goes and asks; --dry-run only prints) - The worktree keeps its identity across sprints — only the work branch changes
In this repository22 other files it gives its agents.
- AGENTS.mdAGENTS.md
- ai-auditSkill
- bulk-readSkill
- canon-slidesSkill
- captureSkill
- context-checkSkill
- context-doctorSkill
- dead-code-cleanupSkill
- doc-auditSkill
- learnings-sweepSkill
- mikadoSkill
- mutation-testSkill
- output-validatorSkill
- promote-learningsSkill
- pr-reviewSkill
- repo-auditSkill
- repo-workflow-auditSkill
- sdlc-auditSkill
- skill-evalSkill
- skill-exportSkill
- sprintSkill
- wrapupSkill
The file
We split this CLAUDE.md into the rules it gives an agent
3 rules. About 0 are common, 2 are distinctive, and 1 is 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.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 10
Prefer reversible actions.
- Distinctive: line 17
When running multiple Claude Code instances in parallel worktrees:
- Very specific: line 19
After a PR merges, reset the tracking branch from inside the worktree: tkt worktree-reset (fetches first, refuses on uncommitted, untracked …
≈ “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
canon: Project-Local Instructions
Canon-specific guidance that applies only in this repo. Global instructions are in ~/.claude/CLAUDE.md.
Agent Design
- One skill, one job. A skill that does two things is two skills waiting to be separated.
- Compose agents from tools + context + prompts — not inheritance hierarchies.
- Request only the tools the current task needs. Bloated tool schemas degrade reasoning.
- Prefer reversible actions. Confirm before irreversible ones (send, delete, publish, deploy).
- Fail loudly — surface ambiguity rather than guessing silently.
- Solve with one agent before building a multi-agent pipeline. Complexity compounds failure.
- Requirements define intent; code defines reality. The gap between them is where hallucination lives.
Worktrees (Parallel Agents)
When running multiple Claude Code instances in parallel worktrees:
- Each worktree uses a long-lived tracking branch (e.g.,
claude-code-1/main-1) - After a PR merges, reset the tracking branch from inside the worktree:
tkt worktree-reset(fetches first, refuses on uncommitted, untracked or unpushed work, shows what goes and asks;--dry-runonly prints) - The worktree keeps its identity across sprints — only the work branch changes
@AGENTS.md
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.