by sito-plugins
rig — an AI Quality Operating System for Claude Code, built on a quality-gated workbench. Give /rig:go a natural-language task: it classifies the task, runs it in an isolated git worktree, and checks the result against a per-task-type acceptance gate — the diff only lands in your working tree on an explicit accept. Machine sensors back the gate deterministically (secret leak, prompt-injection markers, gate tampering, destructive commands, OpenAPI schema drift), with recorded overrides as the escape hatch. Track parallel tasks with board and the read-only cockpit; measure whether the gate actually works with drill (seeded-bug reviewer detection rates) and stats (rubber-stamp detection). Compose recipes, personas, and knowledge as reusable bricks; a deterministic orchestrator (orchestrate.py) runs recipe state machines with parallel verification, cost-tier auto-routing learned from run history, A/B recipe experiments, and token/cost metering. Reaches beyond a live session via a stdlib-only MCP server, a GitHub Action, Slack/Teams notifications, Codex/Cursor native-layer adapters, and a read-only VS Code sidebar. Feedback loops close the gap between accept-time predictions and reality: production-outcome tracing, signed provenance, cross-session instinct learning with decay, and gap prescriptions that draft /rig:forge improvement requests. Japanese writing quality is first-class: an AI-smell catalog with cognitive-rhythm grounding and a deterministic prose-rhythm sensor. v2 adds the organisational layer: a common policy that stacks org -> team -> project and can only ever tighten downstream, role-based permissions over a fixed vocabulary, an approval flow with quorum, separation of duties and freshness binding to the approved commit, time-boxed waivers that turn --force into a named and expiring exception, a hash-chained signed audit ledger where edits and deletions are detectable, and conformance scoring that rolls several repositories up into a per-team view. All of it inert until a repository is bound with `rig-wb govern init`, so solo use is unchanged. v2.1 extends governance from the accept step to any stage of a recipe: a step may declare `actor` (the org role that owns it) and `human_gate` (halt until a qualified person signs off), and an org policy may impose a `stage:<id>` approval rule on a recipe that never asked for one. A parked run persists in `awaiting_approval` and resumes via `orchestrate approve`. v2.2 makes a run legible to somebody who does not already know the recipe: `new` prints the chosen recipe's flow map, each step transition shows position and what is next, and board gains a real denominator plus a whose-move-is-it column. It also measures what had only ever been asserted — `rig-wb wb context` counts what rig's own output costs the parent session, which is the part rig can actually observe.
Claude Code1 Skill