メニュー
Sprint lifecycle: plan, run, review, retro. The top-level orchestrator for autonomous issue resolution. Use for "run a sprint", "plan the sprint", "sprint review", "sprint retro", "/sprint", "/sprint plan", "/sprint review", "/sprint retro", or any variant.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Step back and interrogate the frame instead of executing it. `/why <subject>` GENERATES the sharp questions that should have been asked about a request, plan, artifact, or your own intended action — it does NOT answer them. Use when handed an assignment that smells mis-scoped, when about to loosen a constraint to silence a signal (bump a timeout, widen a threshold, disable a check, narrow scope to the measurable half), when triaging borderline/unfiltered backlog issues, or when a user is frustrated that the obvious framing is wrong. Triggers: /why, /why nsfw.
Build an autonomous sprint skill for a target project. Explores the project's workflows, constraints, and tooling, then produces a tailored sprint skill that lets Claude orchestrate parallel implementation sessions end-to-end. Use when setting up auto-sprint in a new repo, or when adapting sprint orchestration to a project with different workflows. Trigger on "bootstrap sprint", "set up auto-sprint", "build sprint skill for <project>", or "make <project> sprintable".
Audit and garbage-collect a committed Claude-memory store (MEMORY.md + per-fact files). Evidence-based, parallel, then a gated apply. Use when MEMORY.md has grown large, after several sprints, when memory has drifted from CLAUDE.md, or when the user says "/memory-gc", "clean up memory", "audit memory", "the memory is huge", or notices memory diffs/orphans.
Extract actionable lessons from sprint history. Buckets every Claude Code session into the sprint it ran in, builds a deterministic per-session digest (tokens, duration, errors, orchestration signals, evidence), then runs a multi-agent Workflow that classifies each session (orchestrator vs worker), scans for tool errors / autoclassifier rejections / user interventions, summarizes it, aggregates per-sprint, per-10-sprint window, and over time — ending in ranked, concrete actions. Use for "extract sprint lessons", "what can we learn from sprint history", "/sprint-lessons", or a retro deep-dive.
Author, refine, or migrate a doing-it-wrong architectural rule, and harvest recurring mistakes from merged PRs. Routes on its argument — `harvest` (mine PR review comments for recurring author mistakes), `init` (bootstrap the rule engine into a repo), `add` (author a rule), or no arg (overview + "is this a good rule?" triage). Use when the user wants to "write a rule", "add a lint rule", "should we lint for this", "make a rule for X", "migrate a check-*.ts into the rule engine", "harvest rules", "what should we lint for", "reduce review churn", "mine PR comments for rules", or judging whether a proposed invariant is worth a rule.
Post-implementation triage: measure actual diff metrics to determine review depth. Run after implement, before review. Usage: /estimate (in a worktree with changes)
| name | sprint |
| description | Sprint lifecycle: plan, run, review, retro. The top-level orchestrator for autonomous issue resolution. Use for "run a sprint", "plan the sprint", "sprint review", "sprint retro", "/sprint", "/sprint plan", "/sprint review", "/sprint retro", or any variant. |
The sprint lifecycle has four phases. Route based on arguments:
| Input | Action |
|---|---|
/sprint plan | → Read references/plan.md |
/sprint (sprint plan exists) or /sprint <N> (sprint number matching a plan file) | → Auto-chain: read references/run.md and execute run; at wind-down, read references/review.md and execute the release inline; then read references/retro.md and execute the retro inline — all in the same session, no re-invocation |
/sprint run or /sprint run <N> | → Run-only: read references/run.md and execute; mark the sprint-plan header with (RUN ONLY) next to the "Started …" timestamp; stop at wind-down without review/retro |
/sprint <issue-numbers> (multiple numbers, or single number with no matching plan) | → Read references/run.md and run those specific issues (no auto-chain — there's no sprint plan to base review/retro on) |
/sprint (no issues, no plan) | → Offer: "No sprint plan found. Run /sprint plan first, or pass issue numbers." |
/sprint review | → Read references/review.md |
/sprint retro | → Read references/retro.md |
Why auto-chain is the default: a sprint's post-run context (~300k+ tokens) is worth more than a single release cut. Running review and retro in a fresh /sprint review invocation pays a full cache miss to re-read everything you already know. Auto-chain executes review + retro inline in the same session and preserves it.
Disambiguation /sprint <N>: if .claude/sprints/sprint-<N>.md exists, <N> is a sprint number → auto-chain. Otherwise treat <N> as an issue number → run-only with no auto-chain.
Sprints are numbered sequentially. The current sprint number is determined by:
ls .claude/sprints/sprint-*.md | sort -t- -k2 -n | tail -1
If no sprint files exist, start at sprint 10 (we've done ~9 unnumbered sprints).
The sprint number threads through all phases:
.claude/sprints/sprint-N.md.claude/diary/yyyyMMdd.N.mdreferences/mcx-claude.md — session management commandsreferences/plan.md — sprint planning phasereferences/run.md — sprint execution phase (orchestrator prose)references/review.md — release + changelog phasereferences/retro.md — retrospective / diary phasereferences/introspection.md — periodic code-first introspection (sprints ending in 7)references/investigations.md — nerd-snipe gate before impl (flakies, recurring bugs, perf/security findings); load-bearing spawn shape (mcx claude spawn, NOT Agent tool — see #2009)Per-phase logic is defined in .mcx.yaml + .claude/phases/*.ts, not
in run.md. Inspect a phase with mcx phase show <name> or preview its
next action with mcx phase run <name> --dry-run. See docs/phases.md for
the manifest schema.
mcx claude wait, not sleep. wait --timeout is event-driven and interruptible.sprint-{N} branch per sprint, in a worktree at .claude/worktrees/sprint-{N}/. All sprint-meta commits — plan, mid-sprint amendments, run-time edits (timestamps, Excluded section), Results section, retro diary, release commit — accumulate on this branch. A single auto-merge PR is opened at plan time (as draft) and converted to ready at retro. This gives one watchable PR per sprint, replaces the older mix of sprintNN/plan / sprintNN/retro / release/vX.Y.Z short-lived branches, and means the orchestrator never pushes directly to main (which the autoapprover blocks). The worktree name matches the branch (.claude/worktrees/sprint-46/ for sprint 46) so leftover state from a previous sprint never collides. Between-sprint meta-fixes (.claude/skills/**, etc., applied via plan.md Step 1a) use a separate meta/<descriptor> branch since they live outside any sprint. See references/{plan,run,review,retro}.md for the per-phase command sequences.