Ideate Component

"A component earns its place by doing one thing that no other component does."

Brainstorm individual components (skill, agent, command, or hook) for an existing or planned plugin. Produces a concept card that defines the component's responsibility, relationships, and placement within the plugin structure.

This is a read-only skill. It proposes but never creates files.

---

Procedure

Step 1 -- Identify Component Type

• Determine which component type the user wants: skill, agent, command, or hook.
• If not specified, ask: "What type of component? A skill (does work), agent (orchestrates), command (user entry point), or hook (lifecycle event)?"
• Reference references/component-patterns.md for the decision matrix.

Step 2 -- Read Existing Plugin Structure

• If a plugin path is provided, read its plugin.json, list existing skills/agents/commands/hooks.
• Build an inventory of current components and their responsibilities.
• If no plugin path is provided, work from a plugin brief or architecture plan if available.
• Tag source: [CODIGO] for file-derived info, [SUPUESTO] for assumptions.

Step 3 -- Propose Name and Responsibility

• Generate 2-3 name candidates in kebab-case.
• For each candidate, write a one-sentence responsibility statement following the pattern: "This {type} {verb} {object} {constraint}."
• Example: "This skill validates hook configurations against the type-event compatibility matrix."
• Verify the name does not conflict with existing components in the plugin.

Step 4 -- Map Relationships

• Identify which existing components this new component interacts with.
• For skills: which agent orchestrates it? Which movement does it belong to?
• For agents: which skills does it manage? What is its delegation boundary?
• For commands: which skill or agent does it invoke? Is it canonical or an alias?
• For hooks: which event triggers it? What context does it need?
• Draw a simple dependency diagram in text or Mermaid format.

Step 5 -- Identify Conflicts and Gaps

• Check for responsibility overlap with existing components.
• If overlap exists, propose one of: merge, split, or differentiate.
• Check for missing dependencies: does this component need something that does not exist yet?
• If dependencies are missing, list them as additional components to ideate.

Step 6 -- Determine MOAT Depth

• Assess the component's complexity to recommend MOAT asset depth:
  • MINIMAL: SKILL.md only (simple, self-contained components).
  • STANDARD: SKILL.md + references/ (needs reference material).
  • FULL: SKILL.md + references/ + examples/ (complex, benefits from examples).
  • DEEP: SKILL.md + references/ + examples/ + prompts/ (user-facing, needs prompt templates).

Step 7 -- Output Concept Card

• Compile findings into a concept card:

  ## Component Concept Card
  - Type: {skill|agent|command|hook}
  - Name: {recommended-name}
  - Responsibility: {one-sentence}
  - Movement: {movement-name or N/A}
  - Dependencies: {list}
  - Conflicts: {list or "none"}
  - MOAT Depth: {MINIMAL|STANDARD|FULL|DEEP}
  - Tools Needed: {list}
  - Estimated Lines: {range}
  

• Present the card for review. Suggest designing the component with /pqa:design.

---

Quality Criteria

• The responsibility statement is a single sentence with a clear verb and object. [DOC]
• Name candidates are verified unique against existing components. [CODIGO]
• Relationship mapping includes all direct dependencies. [CODIGO]
• MOAT depth recommendation has explicit justification. [DOC]
• Conflict analysis explicitly states whether overlap exists and proposes a resolution (merge, split, or differentiate). [DOC]

Assumptions & Limits

• Read-only. This skill proposes component concepts but never creates files or directories.
• Relationship mapping depends on the existing plugin structure being readable. If the plugin is not yet scaffolded, relationships are inferred from the architecture plan or brief.
• MOAT depth recommendations are guidelines, not constraints. The actual depth may change during the design phase when complexity becomes clearer.
• Cannot assess whether the proposed component is actually useful to users -- only that it is structurally sound and non-overlapping.
• Hook compatibility is checked against the type-event matrix, but the skill does not validate the hook's runtime behavior.

Good vs Bad

Bad concept card:

- Type: skill
- Name: do-stuff
- Responsibility: Does stuff.

Missing: no movement, no dependencies, no conflicts, no MOAT depth, no tools.

Good concept card:

## Component Concept Card
- Type: skill
- Name: validate-config
- Responsibility: Validates plugin configuration files against their JSON schemas.
- Movement: VALIDATE
- Dependencies: validate-manifest (shares plugin.json parsing), audit-security (config may contain secrets)
- Conflicts: None — validate-manifest checks plugin.json only; this covers settings.json and hooks.json
- MOAT Depth: STANDARD (needs reference for config schemas)
- Tools Needed: Read, Glob, Grep
- Estimated Lines: 80-120

Includes: type, descriptive name, precise responsibility, movement placement, dependencies, conflict analysis, MOAT depth with rationale, tool list, line estimate.

Anti-Patterns

1. Proposing a component that duplicates an existing one without acknowledging the overlap.
2. Creating a skill for something that should be a command alias.
3. Designing an agent that manages only one skill (merge into the parent agent).
4. Recommending DEEP MOAT depth for every component regardless of complexity.

Edge Cases

1. User wants to add a component to a plugin that has no architecture plan -- ideate the component in isolation, note the lack of context.
2. The proposed component requires tools not available in the plugin sandbox -- document the constraint and propose alternatives.
3. User wants a hook but the event they need is not compatible with their hook type -- reference the compatibility matrix and suggest alternatives.