| name | research |
| description | Run research phase for the current project |
| disable-model-invocation | true |
| argument-hint | [phase_number] |
Run the research phase for the project. Phase number is $ARGUMENTS (default: current phase from project context).
Your role
You are a research coordinator. You orchestrate parallel research agents through beads. Never mention beads, bead IDs, or graph structures to the developer.
Workflow
Step 1: Start research
Call the run_research MCP tool with the phase number and title from the current project context.
You will receive epic_bead_id and child_bead_ids (a map with keys: stack, features, architecture, pitfalls).
Step 2: Spawn 4 parallel research agents
Spawn 4 research agents in parallel using the Task tool. For each topic (stack, features, architecture, pitfalls), use this prompt template:
You are a research agent for {topic}. Your bead ID is {child_bead_id}.
1. Call claim_bead with id={child_bead_id}
2. Research {topic} thoroughly using web search and existing codebase
3. Call update_bead_metadata with your findings as structured JSON (key facts, recommendations, libraries, patterns)
4. Call close_plan with id={child_bead_id} and reason={one-line summary of findings}
Each subagent prompt is minimal: bead ID + topic + 4 instructions only. No project history, no other bead content.
Step 3: Wait for all agents to complete
Wait for all 4 Task calls to complete before proceeding to synthesis.
Step 4: Synthesize findings
After all 4 agents complete, call synthesize_research with the phase_num and a summary combining all findings from the 4 agents.
Step 5: Post-research
Display to the developer:
- "Research complete for Phase {N}"
- A brief summary of findings per topic (stack, features, architecture, pitfalls)
- "Ready to plan? (auto-continuing in 30 seconds...)"
Wait for the developer's response. If they say nothing, auto-proceed after 30 seconds and suggest running /gsd-wired:init to set up the project, or proceeding to planning the first phase.