Sprint

sprint

Orchestrate one large milestone as a series of brainstorm → plan → execute stages, tracked in a living sprint doc, with the implementation delegated to an executor (codex or mimo) while Claude stays the conductor.

You are the foreman. The executor digs.

What it's for

A milestone too big for a single spec-and-plan — a long, multistage effort that spans sessions and needs several rounds of brainstorming and planning. Instead of one giant plan, you decompose it into stages and drive them one at a time, resuming across sessions from a persistent doc.

Use when:

• A milestone needs multiple brainstorm/plan rounds, not one spec → done.
• Work spans sessions and you need to resume "what stage am I on".
• You want to hand the coding to an executor while steering design yourself.

Not for: a single-spec feature (use superpowers:brainstorming → writing-plans → execute) or a one-off task (use codex:rescue or mimo-code directly).

How it works

• Lean conductor. The main context holds only the sprint doc, current stage, decisions, and open questions. Every technical step runs in an isolated git worktree via a subagent, so diffs and build logs never flood it.
• Per-stage lifecycle: brainstorm a spec → write a plan → isolate a worktree → executor implements → headless /code-review --fix → verify → commit & land → update the doc.
• Two executors. The same orchestration drives either engine; only the Execute step differs:
  • mimo — via the mimo-code plugin (a hard dependency, always present).
  • codex — via the official Codex plugin, an optional runtime probe.
• The chosen engine is recorded in the sprint-doc header so cross-session resume keeps the same engine without re-asking.

Requirements

• mimo-code — hard dependency. sprint/.claude-plugin/plugin.json declares "dependencies": ["mimo-code"], so installing sprint auto-installs mimo-code. The mimo executor is therefore always available.
• Codex — optional. The official Codex plugin openai/codex-plugin-cc (codex@openai-codex) provides codex:rescue + the codex-companion runtime. It lives in an external marketplace, so it can't be a hard dependency — it stays a runtime probe.
• Best with: superpowers for specs & plans (the skill degrades and tells you what's missing).
• A git repository (stages use git worktree).

Usage

Command	Engine
/sprint	Probe: if codex is present, ask (mimo vs codex); otherwise mimo.
/sprint mimo	mimo (model resolved per stage).
/sprint codex	codex.
/sprint mimo <provider/model> [variant]	mimo, pinned to that model for the whole sprint.

Describe the milestone after the command. The skill probes capabilities, decomposes the milestone into stages, and drives them.

The mimo model is resolved automatically before every stage (authenticated providers ∩ model catalogue) — unless you pin one with the <provider/model> [variant] form, in which case the pinned model is reused for every stage.

Skill reference →