sartoris-digital/using-superpowers

<SUBAGENT-STOP> If you were dispatched as a subagent to execute a specific task, skip this skill. </SUBAGENT-STOP> <EXTREMELY-IMPORTANT> If you think there is even a 1% chance a skill might apply to what you are doing, you ABSOLUTELY MUST invoke the skill.

IF A SKILL APPLIES TO YOUR TASK, YOU DO NOT HAVE A CHOICE. YOU MUST USE IT.

This is not negotiable. This is not optional. You cannot rationalize your way out of this. </EXTREMELY-IMPORTANT>

Instruction Priority

Superpowers skills override default system prompt behavior, but user instructions always take precedence:

1. User's explicit instructions (AGENTS.md, direct requests) — highest priority
2. Superpowers skills — override default system behavior where they conflict
3. Default system prompt — lowest priority

If AGENTS.md says "don't use TDD" and a skill says "always use TDD," follow the user's instructions. The user is in control.

How to Access Skills

In Pi: Skills are auto-triggered when their description matches your intent. You can also invoke any skill explicitly via its /skill-name slash prompt. Skill content is delivered to you directly — do not read the SKILL.md file yourself; just follow the instructions when they arrive.

When this skill itself loads, the bootstrap extension has already injected its content into the session, so you have it without doing anything special.

Platform Adaptation

Skills were originally authored against Claude Code tool names. Pi uses lowercase equivalents (read, write, edit, bash, grep, find, subagent, state, …). When a skill mentions a Claude Code primitive, consult references/pi-tools.md for the Pi equivalent.

Other harness tool maps (Copilot, Codex, Gemini) live upstream and are intentionally not bundled in Pi.

Using Skills

The Rule

Invoke relevant or requested skills BEFORE any response or action. Even a 1% chance a skill might apply means that you should invoke the skill to check. If an invoked skill turns out to be wrong for the situation, you don't need to use it.

digraph skill_flow {
    "User message received" [shape=doublecircle];
    "About to start planning?" [shape=doublecircle];
    "Already brainstormed?" [shape=diamond];
    "Invoke brainstorming skill" [shape=box];
    "Might any skill apply?" [shape=diamond];
    "Invoke skill (slash prompt or auto-trigger)" [shape=box];
    "Announce: 'Using [skill] to [purpose]'" [shape=box];
    "Has checklist?" [shape=diamond];
    "Track each item as a todo" [shape=box];
    "Follow skill exactly" [shape=box];
    "Respond (including clarifications)" [shape=doublecircle];

    "About to start planning?" -> "Already brainstormed?";
    "Already brainstormed?" -> "Invoke brainstorming skill" [label="no"];
    "Already brainstormed?" -> "Might any skill apply?" [label="yes"];
    "Invoke brainstorming skill" -> "Might any skill apply?";

    "User message received" -> "Might any skill apply?";
    "Might any skill apply?" -> "Invoke skill (slash prompt or auto-trigger)" [label="yes, even 1%"];
    "Might any skill apply?" -> "Respond (including clarifications)" [label="definitely not"];
    "Invoke skill (slash prompt or auto-trigger)" -> "Announce: 'Using [skill] to [purpose]'";
    "Announce: 'Using [skill] to [purpose]'" -> "Has checklist?";
    "Has checklist?" -> "Track each item as a todo" [label="yes"];
    "Has checklist?" -> "Follow skill exactly" [label="no"];
    "Track each item as a todo" -> "Follow skill exactly";
}

Red Flags

These thoughts mean STOP—you're 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. | | "Let me gather information first" | Skills tell you HOW to gather information. | | "This doesn't need a formal skill" | If a skill exists, use it. | | "I remember this skill" | Skills evolve. Read 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. | | "This feels productive" | Undisciplined action wastes time. Skills prevent this. | | "I know what that means" | Knowing the concept ≠ using the skill. Invoke it. |

Skill Priority

When multiple skills could apply, use this order:

1. Process skills first (brainstorming, debugging) - these determine HOW to approach the task
2. Implementation skills second (frontend-design, mcp-builder) - these guide execution

"Let's build X" → brainstorming first, then implementation skills. "Fix this bug" → debugging first, then domain-specific skills.

Skill Types

Rigid (TDD, debugging): Follow exactly. Don't adapt away discipline.

Flexible (patterns): Adapt principles to context.

The skill itself tells you which.

User Instructions

Instructions say WHAT, not HOW. "Add X" or "Fix Y" doesn't mean skip workflows.

Pi Agents

These agents are available via the subagent extension:

| Agent | Tier | Purpose | |-------|------|---------| | architect | reasoning | Plan review and completion verification | | critic | reasoning | Adversarial quality review of plans and designs | | designer | standard | UI/frontend components, styling, accessibility | | writer | fast | Documentation, READMEs, changelogs | | researcher | standard | External documentation and API reference lookup | | scientist | standard | Data analysis, hypothesis testing | | vision | standard | Visual analysis of screenshots and mockups | | worker | standard | General-capability implementer (Pi's general-purpose equivalent) |

Dispatch with subagent (single, parallel, or chain mode) — see references/pi-tools.md for the JSON shapes.

Tier System

Model tiers map agent names to model capabilities:

| Tier | Default Model | Used For | |------|--------------|----------| | fast | claude-haiku-4-5 | Quick lookups, documentation, recon | | standard | claude-sonnet-4-6 | Implementation, review, analysis | | reasoning | claude-opus-4-6 | Complex debugging, security, architecture |

Override via .pi/superpowers.json:

{
  "models": {
    "fast": "claude-haiku-4-5",
    "standard": "claude-sonnet-4-6",
    "reasoning": "claude-opus-4-6"
  }
}

Pass tiers explicitly in subagent dispatch:

{ "agent": "worker", "task": "...", "tier": "standard" }

Execution Strategies

The executing-plans skill supports 4 strategies:

| Strategy | Keyword | Best For | |----------|---------|----------| | Sequential | (default) | Dependent tasks, small plans | | Parallel | "parallel" | Many independent tasks | | Autopilot | "autopilot" | Large plans, hands-off | | Ralph | "ralph" | PRD with user stories |

Magic Keywords

These keywords are detected automatically by the orchestrator:

| Keyword | Action | |---------|--------| | ralph | Activate ralph (story-driven) execution strategy | | autopilot | Activate autopilot execution strategy | | ecomode | Toggle token-efficient model routing | | cancel | Cancel active execution mode | | ralplan | Start consensus planning | | parallel | Use parallel execution strategy |

Planning Workflows

• /plan — Unified entry point. Detects broad vs. specific requests, routes to interview or direct planning.
• /ralplan — Consensus planning: planner → architect → critic review loop until approved.

Verification Tiers

The verification-before-completion skill dispatches the architect agent at different tiers:

| Tier | When | Evidence Required | |------|------|-------------------| | Light (fast) | <5 files, <100 lines | Build passes + diagnostics clean | | Standard | Default | Build + full test suite passes | | Thorough (reasoning) | >20 files, security, architecture | Full architect review + all tests + security |

State Management

The state extension provides persistent state for execution modes:

state read <key>        — Read state
state write <key> {...} — Write state
state clear <key>       — Delete state
state list              — List all active states

Used by autopilot, ralph, ecomode, and the persistence engine.