Spec-Anchored Agentic Development — methodology + ready-to-install Claude Code bundle: one permanent spec per capability, machine gates before human review, autonomy as an evidence-gated gradient
After implementing, generate a complete reference walkthrough of the changes made — every dimension (what, how, why), anchored in the spec, plan, and diff — as a self-contained interactive HTML page in docs/walkthroughs/.
Orchestrate a whole project (or slice) of GitHub issues into resolution-gated waves of Orca child worktrees — one autonomous worker per ticket — and babysit every PR through CI, the applicable-lens review, and third-party feedback until the human merges. Waves are resolution-gated — merged PRs, corroborated no-change…
Prepare a repository's verification infrastructure for this system — the three-command interface (check / check- / golden), the metric-class gates (stack-agnostic — the class is the requirement, the tool is an instance), the golden harness skeleton, minimum CI, and the ratchet baseline. Brownfield-safe: detect…
Shape work through a relentless interview — interrogate an idea, a transcript, existing code, or an existing spec until shared understanding; or sharpen a task until it's implementable. Works the question frontier in rounds — every currently-askable question at once, numbered, each with a recommended answer — the…
Break a capability spec (or the shaping conversation that produced it, or a parent issue) into tracer-bullet tickets — vertical slices anchored on the spec's pointed stable IDs (typed, never renumbered), each declaring its blocking edges. Quizzes the human on the breakdown, then publishes to a local tickets.md or to…
Write or update the capability spec from an interview — turns a /shape session (or provided notes/transcript) into specs/ / .md on the template. Does not interview back; gaps become open questions (OQ- -###), never guesses. Every BR/AC carries a stable typed ID (BR- -###, AC- -###) issued in continuation; existing IDs…