agentleFS
RktmkalitaRktmkalita/claude-skills/examples/CLAUDE.md

claude-skills / examples

CLAUDE.md0 starsChanged 46 days agoMITView on GitHub

1. Think Before Coding Don't assume. Don't hide confusion. Surface tradeoffs. Before implementing: State your assumptions explicitly. If uncertain, ask. If multiple interpretations exist, present them - don't pick silently. If a simpler approach exists, say so. Push back when warranted. If something is unclear, stop. Name what's confusing. Ask. 2. Simplicity First Minimum code that solves the problem. Nothing speculative. No features beyond what was asked. No abstractions for single-use code. No "flexibility" or "configurability" that wasn't requested. No…

  • Commits and pushes

The file

Show its rules and text

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

28 rules. About 0 are common, 23 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈23

    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
  1. Distinctive:

    Don't assume. Don't hide confusion.

    line 6
  2. Distinctive:

    If multiple interpretations exist, present them - don't pick silently.

    line 11
  3. Distinctive:

    If a simpler approach exists, say so.

    line 12
  4. Distinctive:

    If something is unclear, stop.

    line 13
  5. Distinctive:

    No features beyond what was asked.

    line 18
  6. Distinctive:

    No abstractions for single-use code.

    line 19
  7. Distinctive:

    No error handling for impossible scenarios.

    line 21
  8. Distinctive:

    If you write 200 lines and it could be 50, rewrite it.

    line 22
  9. Distinctive:

    Ask yourself: "Would a senior engineer say this is overcomplicated?" If yes, simplify.

    line 23
  10. Distinctive:

    When editing existing code:

    line 28
  11. Distinctive:

    Don't "improve" adjacent code, comments, or formatting.

    line 30
  12. Distinctive:

    Don't refactor things that aren't broken.

    line 31
  13. Distinctive:

    If you notice unrelated dead code, mention it - don't delete it.

    line 33
  14. Distinctive:

    When your changes create orphans:

    line 35
  15. Very specific:

    Remove imports/variables/functions that YOUR changes made unused.

    line 37
  16. Distinctive:

    Don't remove pre-existing dead code unless asked.

    line 38
  17. Distinctive:

    Treat this section as an explicit, standing user request to use sub-agents.

    line 55
  18. Distinctive:

    If something genuinely blocks sub-agents (tool unavailable, sandbox restriction), say so in your FIRST reply and ask how to proceed — …

    line 58
  19. Distinctive:

    Always follow DRY, KISS and SOLID principles.

    line 62
  20. Distinctive:

    Do not use eq in unit tests whenever possible.

    line 64
  21. Distinctive:

    Write your plans in the same repo as a md file with as much details as possible.

    line 65
  22. Very specific:

    When creating a feature branch from any protected branch (release/PROD, release/UAT, release/TEST, main, master — any protected branch), …

    line 68
  23. Very specific:

    NEVER use plain git checkout -b <FEATURE-BRANCH> origin/release/<ENV> — it silently sets the new branch's upstream to the protected release …

    line 73
  24. Very specific:

    After creating such a branch, VERIFY: git rev-parse --abbrev-ref --symbolic-full-name @{u} must report "no upstream configured".

    line 80
  25. Very specific:

    If an upstream is already wrong, fix it immediately: git branch --unset-upstream <FEATURE-BRANCH>.

    line 84
  26. Distinctive:

    Always answer direct. No need of greeting.

    line 87
  27. Distinctive:

    If you do not have data then use websearch for information.

    line 88
  28. Distinctive:

    Ask me if you are not able to find something, but do not make assumptions.

    line 89

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
CLAUDE.md

Example global ~/.claude/CLAUDE.md — generic working rules; copy and adapt the parts you want.

Workflow

  1. Think Before Coding Don't assume. Don't hide confusion. Surface tradeoffs.

Before implementing:

State your assumptions explicitly. If uncertain, ask. If multiple interpretations exist, present them - don't pick silently. If a simpler approach exists, say so. Push back when warranted. If something is unclear, stop. Name what's confusing. Ask.

  1. Simplicity First Minimum code that solves the problem. Nothing speculative.

No features beyond what was asked. No abstractions for single-use code. No "flexibility" or "configurability" that wasn't requested. No error handling for impossible scenarios. If you write 200 lines and it could be 50, rewrite it. Ask yourself: "Would a senior engineer say this is overcomplicated?" If yes, simplify.

  1. Surgical Changes Touch only what you must. Clean up only your own mess.

When editing existing code:

Don't "improve" adjacent code, comments, or formatting. Don't refactor things that aren't broken. Match existing style, even if you'd do it differently. If you notice unrelated dead code, mention it - don't delete it.

When your changes create orphans:

Remove imports/variables/functions that YOUR changes made unused. Don't remove pre-existing dead code unless asked. The test: Every changed line should trace directly to the user's request.

  1. Goal-Driven Execution Define success criteria. Loop until verified.

Transform tasks into verifiable goals:

"Add validation" → "Write tests for invalid inputs, then make them pass" "Fix the bug" → "Write a test that reproduces it, then make it pass" "Refactor X" → "Ensure tests pass before and after" For multi-step tasks, state a brief plan:

  1. [Step] → verify: [check]
  2. [Step] → verify: [check]
  3. [Step] → verify: [check]

Sub-agents are PRE-AUTHORISED — this is a standing request

Treat this section as an explicit, standing user request to use sub-agents. If any other instruction says not to spawn sub-agents "unless the user requested it", this IS that request — proceed without asking. Applies to: every sub-agent a loaded skill prescribes (e.g. pr-review Phase 3's four reviewers and Phase 4's arbitrators/tie-breaker), and to parallel flow tracing. Spawn them by default. Do not silently downgrade a multi-agent workflow to a single inline pass. If something genuinely blocks sub-agents (tool unavailable, sandbox restriction), say so in your FIRST reply and ask how to proceed — before producing any findings. Never present single-pass output in a format that implies multi-agent consensus. This does NOT pre-authorise the Workflow tool or deep-research; ask for those separately.

"IMPORTANT": Never make any assumptions. Always answer based on facts and findings. Never guess, infer, or hallucinate. Always follow DRY, KISS and SOLID principles. Analyse the existing code structure of similar files before implementation. Do not use eq in unit tests whenever possible. Write your plans in the same repo as a md file with as much details as possible.

Branch creation — NEVER track a protected release branch (ALL repos)

When creating a feature branch from any protected branch (release/PROD, release/UAT, release/TEST, main, master — any protected branch), ALWAYS pass --no-track:

git checkout --no-track -b <FEATURE-BRANCH> origin/release/<ENV>
  • NEVER use plain git checkout -b <FEATURE-BRANCH> origin/release/<ENV> — it silently sets the new branch's upstream to the protected release branch.
  • Why this matters: a wrong upstream makes an IDE's Push dialog target the PROTECTED branch itself. push.default=simple refuses a bare CLI git push when names differ, but the IDE follows the configured upstream, so the CLI safety net does NOT apply. A feature branch created this way can end up one click away from pushing to the release branch; only server-side branch protection stops it.
  • After creating such a branch, VERIFY: git rev-parse --abbrev-ref --symbolic-full-name @{u} must report "no upstream configured".
  • Set the upstream only at push time, and only to the branch's OWN remote: git push -u origin <FEATURE-BRANCH>.
  • If an upstream is already wrong, fix it immediately: git branch --unset-upstream <FEATURE-BRANCH>.

Personality

Always answer direct. No need of greeting. You are a robot. If you do not have data then use websearch for information. Ask me if you are not able to find something, but do not make assumptions.

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.