Sahir619/fable-method

Fable Domain discusses a domain with the user, researches it from real sources, and generates a trusted skill bundle containing a step-by-step workflow with flowchart, a domain adapter, a trap fixture, and a smoke eval. Use it when creating a new domain adapter for the Fable method or giving a lesser model Fable's workflow for a domain.

Fable Judge re-runs claimed verifications, diffs what actually changed, and detects weakened tests, false completion claims, and scope creep to deliver an evidence-based verdict. Use it after an agent or model claims work is complete.

Fable Loop orchestrates a task through planning, execution, and adversarial verification using parallel subagents. Use it for non-trivial multi-step tasks when the user invokes '/fable-loop' or asks for work done in the style of Fable.

Fable Method structures multi-step work into a loop: classify the ask, define done, gather evidence, decide, act surgically, verify by observation, and report outcome-first. Use when the user says "/fable-method", "use the fable method", or "approach this like Fable", or proactively when starting any multi-step task no task-specific skill covers. Supports plan, audit, and report subcommands.

Release Helper runs ship.py after any config.json edit so configuration changes reach production. Use it whenever you modify settings or fix a configuration bug in this project.