agentleFS
L-X-TL-X-T/ng-agentic/.claude/CLAUDE.md

ng-agentic

CLAUDE.md8 starsChanged 43 days agoMITView on GitHub

Follow ../AGENTS.md for the shared project rules. This file owns Claude Code model routing and tool mechanics. Use Claude Fable 5.1 as the session model. Fable handles planning, design, implementation, investigation, and fixes itself. Any implementation subagent must also use Fable 5.1. Do not delegate implementation or investigation to Opus or Astra, and do not use the Codex CLI for those tasks. This routing applies only to the Claude Code harness. When the user asks for a review, run three…

1 other project uses a version of this.

Adopted by

Most recently taken up by

  • Other versions1 project uses a version of this.

    Word for word · 1

    1. L-X-T/ng-agentic-skills★ 13100% same

    Edited · 0

    None found.

    Wrote their own version · 0

    None found.

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository3 other files it gives its agents.

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
  1. Very specific:

    Follow ../AGENTS.md for the shared project rules.

    line 3
  2. Distinctive:

    Use Claude Fable 5.1 as the session model.

    line 8
  3. Distinctive:

    Do not delegate implementation or investigation to Opus or Astra, and do not use the Codex CLI for those tasks.

    line 10
  4. Distinctive:

    When the user asks for a review, run three independent reviewers in parallel:

    line 15
  5. Distinctive:

    Run Astra directly from Bash, with the model and read-only sandbox explicitly selected:

    line 32
  6. Distinctive:

    Use a unique report file and run the command in the background alongside the two Claude reviewers.

    line 43
  7. Very specific:

    Every mention of a local port or server in a reply is a full clickable URL (http://localhost:4200/register), never a bare :4200 or 4200.

    line 55

≈ “Common” is an estimate: at least 1 in 5 of the 27,641 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

Claude Instructions

Follow ../AGENTS.md for the shared project rules. This file owns Claude Code model routing and tool mechanics.

Model routing – Claude Code

Use Claude Fable 5.1 as the session model. Fable handles planning, design, implementation, investigation, and fixes itself. Any implementation subagent must also use Fable 5.1. Do not delegate implementation or investigation to Opus or Astra, and do not use the Codex CLI for those tasks. This routing applies only to the Claude Code harness.

Reviews when requested

When the user asks for a review, run three independent reviewers in parallel:

  • Claude Fable 5.1, via the Agent tool with model: 'fable'.
  • Claude Opus 5.5, via the Agent tool with model: 'opus'.
  • GPT-6 Astra, via the Codex CLI in read-only mode as described below.

Give all three the same diff, scope, project rules, and review instructions. They report findings only; they must not edit files or launch further agents. For each finding require severity (must/should/nit), file:line, a concrete defect, and a failure scenario.

Fable merges and deduplicates the reports and verifies findings against the code before reporting them. A review request does not authorize applying its findings. When the user asks for fixes, Fable implements them. Normal implementation still runs the project's required checks, but does not automatically launch this three-model review.

Astra review mechanics

Run Astra directly from Bash, with the model and read-only sandbox explicitly selected:

codex exec -m gpt-6-astra -s read-only -C <repo-root> -o <unique-report-file-in-/tmp> -

Pass the self-contained review prompt through stdin. Include the repository root, absolute paths, the same review instructions as the Claude reviewers, and exactly which diff to review: uncommitted changes, a branch comparison, or a specified commit. Embed the diff when it cannot be recovered from that checkout.

Use a unique report file and run the command in the background alongside the two Claude reviewers. Read the final report when it finishes. If a wrapper or connection fails, check whether the Codex process is still running before retrying. Do not rely on the global Codex model default or the global codex-astra wrapper, whose model and execution mode can differ. Never use --dangerously-bypass-approvals-and-sandbox.

The Astra review prompt must forbid file edits, fixes, Git state changes, dev servers, and package installs. It should inspect the supplied diff and report findings, not run formatting or implementation commands.

Ports are links – hard rule

Every mention of a local port or server in a reply is a full clickable URL (http://localhost:4200/register), never a bare :4200 or 4200. Applies to every mention, in every reply, including status lines and summaries. See the dev-server rules in ../AGENTS.md.

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.