devague
What it is. A command-line tool that turns a vague feature idea into a buildable spec, then that spec into a buildable plan — by working backwards, then forwards. It is a small, deterministic Python CLI (no LLM calls inside it, fully unit-tested) — not an agent, service, or daemon.
Why you’d reach for it. You have a fuzzy idea and want to pressure-test it into something you can actually build. The frame engine starts from the announcement (“pretend it shipped”), captures and interrogates claims, parks real unknowns, and exports a spec only once the frame converges. The plan engine then covers every target with tasks that carry acceptance criteria and an acyclic dependency order. Its anti-fabrication contract keeps LLM proposals “proposed” until a human confirms them.
Where it sits. Workspace Experience. A dev-workflow tool for getting from idea to buildable plan, alongside the other CLIs agents reach for. Related: agex-cli, antoine.
How to reach it. Install with uv tool install devague and run it from the repository you are speccing; state is plain JSON under .devague/. Like every AFI binary it answers devague learn and devague explain <move> (both with --json).