Brainstormer Agent
Your technical advisor who challenges assumptions, debates approaches, and provides brutally honest assessments before code is written.
When to Use
- Evaluating multiple architectural approaches with competing trade-offs
- Challenging assumptions about requirements and feasibility
- Debating technical decisions before committing resources
- Need a “second opinion” on complex problems requiring creative solutions
Key Capabilities
| Capability | Description |
|---|---|
| Multi-approach analysis | Explores 3-5 valid solutions with detailed pros/cons |
| Assumption challenging | Questions requirements to ensure solving real problems |
| Trade-off evaluation | Assesses complexity, maintenance, performance, cost, technical debt |
| Brutal honesty | Unfiltered assessment: overengineering alerts, risk identification, simpler alternatives |
| YAGNI/KISS/DRY | Applies principles to prevent premature optimization |
| Contextual recommendations | Considers team skills, timeline, budget constraints |
Common Use Cases
Startup CTO: Evaluating API architecture
/brainstorm [should we use REST API or GraphQL for our mobile app? Team has REST experience, need to ship MVP in 6 weeks]Tech Lead: Debating caching strategies
/brainstorm [API responses are slow. Evaluating Redis cache vs database optimization vs CDN. Current p95 is 800ms, need <200ms]Product Engineer: Challenging feature requests
/brainstorm [stakeholder wants real-time dashboard with WebSockets. Is this necessary or can we use polling?]Solo Developer: Planning refactoring
/brainstorm [monolith is getting messy. Should I split into microservices or improve modular monolith structure first?]Team Lead: Assessing feasibility
/brainstorm [client wants offline-first mobile app with sync. Team has 3 junior devs, 8-week timeline. Realistic?]Pro Tips
Provide context upfront: Team skills, timeline, budget, constraints. Better context = better recommendations.
Be open to challenges: Brainstormer will question your assumptions. That’s the point. Saves costly mistakes.
Focus on “Recommendation” section: If overwhelmed by options, read the final recommendation first, then backtrack for details.
Use success criteria: Brainstormer defines measurable criteria (e.g., “p95 <200ms”). Use these to validate decisions later.
Don’t skip “Open Questions”: Unresolved questions in the output need answers before proceeding.
Trust YAGNI assessments: If brainstormer says it’s overengineering, it probably is. Build what you need now.
Follow “Next Steps”: Output includes actionable next steps. Usually involves /plan command for implementation.
Related Agents
- Planner Agent - Creates implementation plans after approach is decided
- Researcher Agent - Provides data for brainstorming decisions
- Docs-Manager Agent - Documents architectural decision records
Key Takeaway
Brainstormer doesn’t write code. It saves you from writing the wrong code. Use it before committing to complex architectural decisions. 10 minutes of brainstorming prevents weeks of refactoring.