| name | orchestrate |
| description | Coordinate multi-agent workflow for complex tasks |
| user-invocable | true |
/orchestrate
Run a coordinated multi-agent workflow for complex tasks that benefit from
specialized perspectives.
Arguments
$ARGUMENTS — Description of the task to orchestrate
Workflow
1. Planning Phase
Delegate to the planner agent:
- Break down the task into discrete, ordered subtasks
- Identify which agent is best suited for each subtask
- Present the plan for user approval
2. Execution Phase
For each subtask in order:
- Delegate to the appropriate agent
- Collect the agent's output
- Verify the output meets acceptance criteria
- If issues found, iterate with the agent or escalate
3. Integration Phase
After all subtasks complete:
- Run the code-reviewer agent on all changes
- Run the security-reviewer agent if security-sensitive
- Consolidate findings and present summary
4. Verification Phase
Run the /verify command to ensure:
- All tests pass
- No type errors
- No lint violations
- Security checks clean
Agent Roster
| Agent | Use For |
|---|
| planner | Breaking down requirements |
| architect | System design decisions |
| tdd-guide | Test-first implementation |
| code-reviewer | Quality review |
| security-reviewer | Security audit |
| database-reviewer | Schema/query review |
| e2e-runner | E2E test creation |
| doc-updater | Documentation sync |
Output
Present a summary:
- Tasks completed: X/Y
- Agents used: [list]
- Issues found: [count by severity]
- Final verdict: READY / NEEDS WORK