By bapttiste73
Plan with Fable 5, execute with Sonnet. Adds /fableplan to toggle an opusplan-style dual-model configuration.
Plan with Fable 5, execute with Sonnet — an opusplan-style dual-model setup for Claude Code.
Claude Code ships with a built-in opusplan model setting: "Opus in plan mode, Sonnet otherwise". There is no Fable equivalent in the model picker — the model list is hardcoded in the CLI. This plugin recreates it with a small trick: opusplan resolves its plan-mode model through the ANTHROPIC_DEFAULT_OPUS_MODEL environment variable, so pointing that variable at Fable 5 gives you Fable 5 in plan mode, Sonnet for execution.
/plugin marketplace add bapttiste73/fableplan
/plugin install fableplan@fableplan
| Command | Effect |
|---|---|
/fableplan | Enable: plan with claude-fable-5, execute with Sonnet |
/fableplan 1m | Same, with the 1M-context Fable variant (claude-fable-5[1m]) |
/fableplan off | Revert to your previous model configuration |
Restart your Claude Code session after toggling — the model setting is read at startup.
The skill merges two keys into your global ~/.claude/settings.json (everything else is preserved):
{
"model": "opusplan",
"env": {
"ANTHROPIC_DEFAULT_OPUS_MODEL": "claude-fable-5"
}
}
That's it. You can also apply this by hand without installing anything.
ANTHROPIC_DEFAULT_OPUS_MODEL is an undocumented override and could change in a future CLI release./fableplan off.MIT
Based on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimnpx claudepluginhub bapttiste73/fableplan --plugin fableplanGet a second opinion from Claude Fable 5 before big decisions.
Turns Fable 5 into a token-frugal orchestrator: Fable keeps judgment, tiered Opus/Sonnet/Haiku agents do the labor.
Fable 5's working loop for hard tasks: decompose by verification boundaries, verify against the world, choose next by plan-change. Includes scout + refuter agents and a preflight script.
A harness that makes Opus (or any Claude model) behave like Fable. It enforces completion, evidence, and verification as procedure, and auto-routes the right verified pack per task: render-output verification, a multi-story evidence gate, an investigation protocol, and an early-stop guard. It does not fake model capability — see README for the full analysis of what transfers and what does not.
Delegates implementation planning to a Fable 5 planning subagent, then builds from the vetted plan — posting it to a referenced GitHub issue along the way.
The Fable Workflow: how Claude Fable 5 worked, distilled into four skills any model can run. Think (fable-method), act (fable-loop), prove (fable-judge), grow (fable-domain). Evidence-backed: ships its own eval, failures included.