Using OPM

This file is injected at SessionStart by the OPM plugin hook, so it stays short. Subagents dispatched to execute a single task: ignore this skill.

The Rule

If there is even a small chance a skill applies to what you are about to do, invoke it BEFORE any other action: before clarifying questions, before exploring the codebase, before checking files. If it turns out wrong for the situation, you do not have to use it.

Then announce "Using opm:<name> to <purpose>" and follow the skill exactly. If it has a checklist, create a todo per item.

How to Invoke

• Skill tool: Skill(skill: "opm:<name>"), or the slash command /opm:<name>.
• Process skills come first (they set the approach); implementation skills carry it out.
• "Run this plan" -> opm:executing-plans.
• Before entering plan mode: if you have not brainstormed yet, invoke opm:brainstorming first.

Pick the size first

Size	Signal	Skill
spike	one file or about 10 lines, a clear bug or tweak	opm:tdd-workflow, no spec, plan or reviewer
bounded	a feature of a few tasks, design needed	opm:brainstorming then opm:writing-plans
multi-week	more than about a week or 8 tasks	opm:milestone-planning
unattended	a whole new project from one brief	opm:jump-start <project-name> <brief>
contested	the idea itself is unsettled	opm:brew-idea <brief>

When no row fits, ask one question about size.

Scale the approach

Use Claude Code's power without being asked; announce each step in one line ("Using 3 parallel subagents to ...").

• Wide search, review or audit -> a subagent, so the main context stays small.
• Independent pieces -> parallel subagents in one message; git worktrees when they edit files.
• More than about 3 tasks -> opm:writing-plans; weeks of work -> opm:milestone-planning.
• Many agents at once -> offer a multi-agent Workflow with a rough agent count; run it only after the user says yes.

Read references/orchestration.md when you are about to escalate.

Workflow Map

1. opm:brainstorming - design, get approval. Architectural work writes docs/specs/YYYY-MM-DD-<topic>.md.
2. opm:writing-plans - turn the approved spec into bite-sized TDD tasks in docs/plans/YYYY-MM-DD-<topic>.md.
3. opm:executing-plans - fresh subagent per task, reviewer gate, progress ledger. Every task follows opm:tdd-workflow.
4. opm:verification-before-completion - fresh evidence before any "done", commit, or PR.
5. opm:compound-learnings - capture anything non-obvious learned along the way.

Optional: /opm-video:explainer-video <path> turns this into a narrated explainer (install the opm-video plugin first).

Red Flags

These thoughts mean STOP - you are rationalizing:

Thought	Reality
"This is just a simple question"	Questions are tasks. Check for skills.
"I need more context first"	Skill check comes BEFORE clarifying questions.
"Let me explore the codebase first"	Skills tell you HOW to explore. Check first.
"I can check git/files quickly"	Files lack conversation context. Check for skills.
"This doesn't need a formal skill"	If a skill exists, use it.
"I remember this skill"	Skills evolve. Read the current version.
"This doesn't count as a task"	Action = task. Check for skills.
"The skill is overkill"	Simple things become complex. Use it.
"I'll just do this one thing first"	Check BEFORE doing anything.
"I know what that means"	Knowing the concept is not using the skill. Invoke it.

Precedence

User instructions (CLAUDE.md, rules files, direct requests) override skills; skills override default behavior. Skip a skill workflow only when the user has explicitly told you to.

<!-- Adapted from obra/superpowers (MIT) -->