agentleFS
wickedhardflipwickedhardflip/claude-code-tips/skills/fable-setup/templates/CLAUDE.md

claude-code-tips / templates

CLAUDE.md1 starsChanged 38 days agoMITView on GitHub

Main model: Fable 5.2. Subagents default to Opus (set in .claude/settings.json). Fable is the main model in this folder by choice, so any parent CLAUDE.md rule that limits Fable to planning does not apply here. {{REPO_LINE}} Fable plans. Opus builds. Sonnet fixes. Haiku fetches. - Fable (main) is the lead. It plans, makes every judgment call, reviews every diff, commits, and writes the summaries. Edits that need judgment stay on the main model, never on a cheap one. - Use…

The file

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

2 rules. About 0 are common, 1 is distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    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
  1. Distinctive:

    When a task splits into independent parts, run up to 8 agents at once in one message, in the background.

    line 19
  2. Very specific:

    Never print secrets. Redact with [REDACTED].

    line 39

≈ “Common” is an estimate: at least 1 in 5 of the 28,029 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

{{NAME}} - Fable workspace (created {{DATE}})

Main model: Fable 5.2. Subagents default to Opus (set in .claude/settings.json). Fable is the main model in this folder by choice, so any parent CLAUDE.md rule that limits Fable to planning does not apply here.

{{REPO_LINE}}

How to work here

Fable plans. Opus builds. Sonnet fixes. Haiku fetches.

  • Fable (main) is the lead. It plans, makes every judgment call, reviews every diff, commits, and writes the summaries. Edits that need judgment stay on the main model, never on a cheap one.
  • Use the subagents in .claude/agents/. Each one sets its own model::
    • fetcher (Haiku): reading logs, finding files, "where is X used". Read only.
    • fixer (Sonnet): scoped fixes that are already decided, and writing tests.
    • builder (Opus): planned multi-file work and deeper discovery or review of an area.
    • Any other subagent defaults to Opus (CLAUDE_CODE_SUBAGENT_MODEL), never Fable.
  • When a task splits into independent parts, run up to 8 agents at once in one message, in the background. Say which agent and model each one is before spawning.
  • Brief each agent so it can work without your context: the goal, the exact files or area, what is already ruled out, the limits below, and the output file path. Agents write details to a file under findings/ and return a summary of at most 15 lines.
  • Agents that edit at the same time get separate file scopes. They never build or run tests in a shared tree. The lead builds and tests after they finish, then reviews each diff before committing.
  • Check agent claims before acting on them. Spot-check the evidence for anything high-severity or surprising.
  • Plan mode before touching any file on a big job.
  • Effort starts at medium. Go to high only when it is struggling, then drop back.
  • /compact when the context gets heavy, /clear between jobs.

Read first

Memory is keyed to the folder, so this folder starts with none. Your main memory is in <path to your main project's memory folder>. Read its MEMORY.md, then open the topic files that match the task.

Standing rules

  • Facts only. Never invent a name, role, or cause. Label anything inferred. Identifiers stay verbatim.
  • Never print secrets. Redact with [REDACTED].
  • Git: git -C with full paths. Stage explicit paths, never git add -A or git add .. Never checkout, stash, or reset --hard to tidy up another session's work.
  • Changes to live systems (prod, databases, servers) stay with the lead, in plain view, and only when asked for in this session.

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.