brainstorm

name	brainstorm
description	Explore possibilities for a topic by combining research digests with project knowledge and governance context.

Brainstorm

Usage

/brainstorm <topic>

Overview

Creative exploration tool. Reads research digests, project knowledge, and governance docs to explore possibilities for a given topic. Output stays in the conversation unless the user asks to save it.

Lifecycle position: Before review-issue — exploratory thinking before an issue exists. Can also be used mid-task to explore alternatives.

Steps

1. Load context

Read available knowledge sources:

• .agent/knowledge/research_digest.md — workspace research (if exists)
• .agent/project_knowledge/research_digest.md — project research (symlink to manifest repo's .agents/workspace-context/; may not exist)
• .agent/project_knowledge/ — aggregated project knowledge (symlink to .agents/workspace-context/; may not exist)
• docs/PRINCIPLES.md — workspace principles (for feasibility checks)
• AGENTS.md — workspace-wide agent rules and conventions
• ARCHITECTURE.md — system design and layering constraints
• docs/decisions/ — ADRs for historical context
• Project-level equivalents (e.g., project PRINCIPLES.md, .agents/README.md, ARCHITECTURE.md, docs/decisions/) when the topic is project-specific

Check digest freshness — if older than 30 days, note it and suggest running /research --refresh but don't block on it.

2. Explore the topic

For the given topic, consider:

• What exists: What does the workspace/project already have that relates?
• What's possible: What approaches could work? Draw on research digest entries for external patterns and best practices.
• What's missing: What gaps exist? What would need to be built or changed?
• Trade-offs: What are the pros and cons of different approaches?
• Principle alignment: Would any approach conflict with workspace or project principles?

3. Present findings

Structure the output as:

## Brainstorm: <topic>

### Current State
<what exists today>

### Possibilities
1. **<approach>** — <description>
   - Pros: ...
   - Cons: ...
   - Research: <reference to digest entry if applicable>

2. **<approach>** — ...

### Gaps
- <what's missing that any approach would need>

### Principle Considerations
- <any principle alignment issues>

### Suggested Next Steps
- <concrete actions to explore further or begin implementation>

Guidelines

• Explore, don't decide — present options with trade-offs. The user decides which direction to go.
• Reference research — when drawing on digest entries, cite them. This validates the digests' usefulness and helps the user trace your reasoning.
• Stay grounded — possibilities should be feasible within the workspace's architecture and constraints. Wild ideas are welcome but should be flagged as speculative.
• Don't create issues — brainstorming output may lead to issues, but the user should make that call.
• Keep it conversational — this is the most informal of the lifecycle skills. Be thorough but not rigid.