Lean agentic dev loop for Claude Code: prime → lean-doc → orchestrate → repeat. Eleven standalone skills + an opt-in /flow conductor + a /council decision aid. No hooks, agent-free core — drop in and adapt.
Skill Claude CodeCodex
Part of lean-flow
Use when someone asks to run Storm Research, use the storm-research skill, run the STORM method on a topic, says "storm research this" / "storm report on X" / "give me a STORM briefing on X", or wants a multi-perspective, citation-verified HTML research briefing on a topic. Runs a 4-phase pipeline: five expert lenses…
Skill Claude CodeCodex
Part of eval-judgement-fixture-plugin
Use when starting, resuming, or completing any development task or sprint. Drives a gate-driven loop — quick, mvp, and sprint-bulk modes — with a G1 Scope gate and a G2 Design gate before any commit. Self-contained, no specialist agents.
Skill Claude CodeCodex
Part of lean-flow
Fixture skill, not a real one.
Skill Claude CodeCodex
Part of lean-flow
Run a high-stakes, hard-to-reverse, or ambiguous decision through a council of 5 AI advisors who independently analyze it, peer-review each other anonymously, and synthesize a final verdict to a lean verdict- .md. The opt-in decision aid for genuinely hard forks — the pressure-test before an ADR (STANDARD §4) or a G2…
Skill Claude CodeCodex
Part of lean-flow
Use when debugging a defect, unexpected behaviour, failing test, or performance regression. Runs a disciplined 6-phase loop — build a feedback loop, reproduce, hypothesise, instrument, fix with regression test, cleanup + post-mortem. Do not use for architectural analysis or planning — use /orchestrator mvp instead.
Skill Claude CodeCodex
Part of lean-flow
Conduct the full lean-flow loop end-to-end: assess context, ensure a backlog and sprint exist, build through the gates with the right per-task technique, then close with governance. The opt-in conductor sequences the standalone skills as stages and enforces the gates/governance, but never auto-approves a human gate.…
Skill Claude CodeCodex
Part of lean-flow
Use when ending a session mid-work and a fresh agent (or future you) needs to pick it up — context budget running out, a task blocked, or a deliberate stop. Compacts the live conversation into a handoff document saved to the OS temp dir, with a suggested-skills section and references to durable artifacts. Do not use…
Skill Claude CodeCodex
Part of lean-flow
Use to capture a friction, surprise, or insight as a learning the MOMENT you hit it — mid-work, in quick-mode, or any time a Sprint-Close Retro is too far off. Appends an L-NNN candidate to docs/LEARNINGS.md, or bumps a matching entry's count (which feeds the count≥2 promotion). Complements the Sprint-Close Retro…
Skill Claude CodeCodex
Part of lean-flow
Use when creating, updating, or reviewing technical documentation, ADRs, sprint files, or AI-context files; running the sprint lifecycle (promote / execute / close); or migrating existing docs (dev-flow / adlc-flow / any layout) into the lean-flow standard. Follows the LEAN DOCUMENTATION STANDARD — WHY and WHERE only…
Skill Claude CodeCodex
Part of lean-flow
Use when starting, resuming, or completing any development task or sprint. Drives a gate-driven loop — quick, mvp, and sprint-bulk modes — with a G1 Scope gate and a G2 Design gate before any commit. Self-contained, no specialist agents. Do not use for debugging — use /diagnose; or for converting raw intent into tasks…
Skill Claude CodeCodex
Part of lean-flow
Use when starting a new session and want to load project context in a deterministic order — CLAUDE.md, CONTEXT.md, README, MEMORY index, active sprint, ARCHITECTURE map. Emits a health check showing which files were found versus missing. Everything degrades gracefully; nothing aborts. Do not use for mid-session…
Skill Claude CodeCodex
Part of lean-flow
Use when a design question needs to be felt before committing — sanity-check a state machine / data model / API shape, or mock up a few UI options. Builds throwaway code that answers ONE question, then captures the answer and is deleted. Routes to a terminal logic prototype or web UI variants. Do not use to build the…
Skill Claude CodeCodex
Part of lean-flow
Use when code has become hard to change — surface deepening opportunities that turn shallow modules into deep ones (small interface, lots of behaviour) for testability and AI-navigability. Informed by the domain glossary (CONTEXT.md) and ADRs. Do not use to debug a failure (use /diagnose) or to build new behaviour…
Skill Claude CodeCodex
Part of lean-flow
Use when releasing a patch on any project — auto-detects the manifest (plugin / npm / python / cargo / go / flat), or falls back to a changelog-only entry when no manifest exists; bumps the PATCH version, prepends a CHANGELOG entry, then HARD STOPS before push. Skips the bump entirely if only docs changed. Never runs…
Skill Claude CodeCodex
Part of lean-flow
Use when converting a freeform feature request, ticket URL, PRD, or epic into structured TASK-NNN entries in TODO.md. Enforces an assumption registry, risk scoring, vertical-slice granularity, and validation before writing. Self-contained, no specialist agents. Do not use when a task already exists and is ready to…
Skill Claude CodeCodex
Part of lean-flow
Use when building NEW testable behavior test-first — feature work, business logic, an API surface, a tool wrapper. Runs behaviour-driven red-green-refactor in vertical slices (tracer bullet, then one test → one implementation → repeat). Do not use to debug an existing failure or fix a bug — use /diagnose (it already…
Skill Claude CodeCodex
Part of lean-flow
Use when the existing backlog needs grooming — re-prioritising tasks, flagging stale/duplicate/conflicting items, marking what is ready vs blocked vs needs-info, and routing rejected work out of scope. Operates on TODO.md Backlog only, never the Active Sprint. Do not use to create new tasks (use /task-decomposer) or…
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: