BigHugger
GH Repository · Sahir619

fable-method

The Fable Workflow: how Claude Fable 5 worked, distilled into skills any model can run, with the eval that keeps it honest. Think / act / prove.

stars
2,288
30-day movement
+83/day
Related entries
64
Connections
2
skill-packPythonevaluationfableclaude-mdllmclaude-codeclaudecoding-agentai-agentsagent-skills

A skill pack distilling the 'Fable Workflow' (think / act / prove) into skills that any model can run, shipped as an AGENTS.md file. It includes an evaluation component intended to keep the workflow honest.

You want a reusable, model-agnostic coding-agent workflow with a built-in eval rather than ad-hoc prompting.

Use it to

  • Install the skill pack into a coding agent
  • Apply the think/act/prove workflow to tasks
  • Run the included eval to check agent behavior
  • Adapt the AGENTS.md for your own repo

For Developers building or configuring LLM coding agents

Role
skill-pack
Language
Python
Licence
MIT
Forks
327
Open issues
3
Last push
2026-07-15
Latest release
v1.2.0 · 2026-07-09
Skills shipped
4
topicsagent-skillscoding-agentevaluationclaudellmworkflow