Prompt file imported from X0RSH1FT/Warchive (
.github/prompts/feature-idea-roundtable.prompt.md). Copyright stays with the author.
Use this prompt when you want a guided multi-specialist discussion to generate and rank practical feature ideas before committing to implementation.
Workflow:
- Identify one concrete brainstorming anchor: a product surface, workflow pain point, subsystem, doc area, or question such as "what should this UI need next?"
- Retrieve only enough repository context to scope the discussion and define the first validation boundary for any follow-up execution.
- If scope, audience, or constraints are unclear, use
#askQuestionsfirst. Summarize the proposed roundtable pass in plain language, then ask only for missing decisions that change ideation direction. - Build a short
#todoplan for the roundtable stages:- framing and constraints
- specialist idea generation
- cross-specialist challenge
- shortlist and prioritization
- next-sprint recommendation
- Select 3 to 6 relevant specialists with distinct lenses. Keep the roster narrow and outcome-oriented.
- Dispatch each selected specialist with a clear lens and bounded output contract:
- idea proposals (2 to 4 each)
- expected user or repository impact
- implementation complexity estimate
- main dependency or risk
- Run one challenge pass where each specialist critiques at least one idea from another specialist and proposes a refinement.
- Synthesize all candidate ideas into a reasoning scaffold:
- clustered themes
- duplicate merge decisions
- ranking criteria (value, effort, risk, reversibility, strategic fit)
- prioritized shortlist with confidence and unknowns
- Recommend a next sprint slice using a planner-executor shape:
- first planning task owner
- first execution owner
- first validation step
- Keep
Coordinator Agentin orchestration mode only. Do not perform specialist implementation, testing, documentation, or deep domain work inline. - Route to the next owner when requested:
Planner Agentfor scope slicing,Implementation Agentfor coding,Interface Design Agentfor UI direction,Documentation Agentfor durable write-up,Testing Agentfor validation planning, orReviewer Agentfor signoff. - Validate that all referenced agents and target paths exist before concluding.
Specialist selection guidance:
- Use
Interface Design Agentwhen ideas depend on screen hierarchy, flows, or usability. - Use
Planner Agentwhen narrowing ideas into sprint-ready slices and acceptance criteria. - Use
Systems Architect Agentwhen ideas change boundaries, interfaces, or architecture. - Use
Implementation Agentwhen feasibility, migration effort, or technical debt are central. - Use
Testing Agentwhen validation cost, reliability, or regression exposure drive prioritization. - Use
Documentation Agentwhen discoverability, guidance quality, or knowledge ownership matter. - Use
Creative Philosopher AgentorArtistic Director Agentwhen ideation quality depends on stronger creative divergence.
Few-shot examples:
- Good anchor: What should our repository UI workflow support next for faster sprint triage?
- Good specialist spread:
Interface Design Agent,Planner Agent,Implementation Agent, andTesting Agent. - Good synthesized idea: Add a lightweight sprint-planning dashboard prompt plus checklist sync, ranked high value and medium effort with low migration risk.
- Bad anchor: Give me random ideas for everything.
- Bad orchestration: Coordinator generates ideas alone without delegating to specialists.
Close-out requirements:
- Summarize the brainstorming anchor and selected specialist roster.
- Provide the prioritized shortlist with scoring rationale.
- Identify the recommended next-sprint candidate and why it won.
- Name the immediate next owner and first validation step.
- State whether plan-derived work is exhausted.
- If work remains, name the next highest-priority planned slice.
- Label any extra idea as a suggestion outside the plan.
- Include a concise, imperative commit message when resulting changes are ready to keep.