Documentation

epiphany-brief

epiphany-brief is the first stage of the authoring pipeline brief → spec → plan → executor. It captures a messy brain dump with zero information loss and produces a human-readable living brief plus a machine `handoff.json` you hand to `/epi

epiphany-executor

epiphany-executor is the fourth and terminal stage of the pipeline. It takes a structured plan from epiphany-plan and executes it step by step --- bounded, verified, checkpointed, and recoverable.

epiphany-plan

epiphany-plan is the third stage of the pipeline. It takes a specification from epiphany-spec and decomposes it into a structured execution plan --- either Markdown (human-readable, default) or JSON (machine-strict, for programmatic executo

epiphany-spec

epiphany-spec (v2.1.0) is the second stage of the pipeline. It takes a raw or enhanced input (typically a living brief from epiphany-brief) and produces a versioned, audited specification document with human review gates.

goatcs-harness

goatcs-harness is a tier-3 LLM-as-runtime skill-execution harness. It treats the AI model as the runtime itself: skills are declared as directed graphs (`graph.json`) with typed nodes, edges, gates, and machine-enforced routing. The AI reas