| name | omo-sp-omo-bridge |
| description | Bridge between Superpowers workflow roles and oh-my-opencode agents. Automatically maps SP roles to OmO agents. Loads before any skill dispatches sub-agents. |
SP → OMO Agent Bridge
When Superpowers workflow dispatches a sub-agent role, use this mapping to route to the correct oh-my-opencode agent.
Role → Agent Mapping
| SP Role | OMO Tool | Parameters |
|---|
| Implementer (simple) | task() | category="quick", load_skills=["omo-test-driven-development"] |
| Implementer (complex) | task() | category="deep", load_skills=["omo-test-driven-development"] |
| Implementer (hard logic) | task() | category="ultrabrain", load_skills=["omo-test-driven-development"] |
| Implementer (frontend) | task() | category="visual-engineering", load_skills=["omo-test-driven-development", "frontend-ui-ux"] |
| Spec Reviewer | task() | subagent_type="momus" |
| Code Quality Reviewer | task() | subagent_type="oracle" |
| Codebase Search | task() | subagent_type="explore", run_in_background=true |
| External Research | task() | subagent_type="librarian", run_in_background=true |
| Pre-planning Analysis | task() | subagent_type="metis" |
| Plan Creation | task() | subagent_type="prometheus" or /start-work command |
| Plan Review | task() | subagent_type="momus" |
Automatic Skill Injection
When dispatching implementation tasks, ALWAYS inject relevant skills:
| Task Type | load_skills |
|---|
| Any implementation | ["omo-test-driven-development"] |
| Bug fix | ["omo-test-driven-development", "omo-systematic-debugging"] |
| Feature work | ["omo-test-driven-development", "omo-verification-before-completion"] |
| Frontend | ["omo-test-driven-development", "frontend-ui-ux"] |
| Git operations | ["git-master"] |
Two-Stage Review (Superpowers Discipline)
When executing subagent-driven-development workflow:
- After implementer DONE → dispatch
subagent_type="momus" with spec compliance prompt
- After spec OKAY → dispatch
subagent_type="oracle" with code quality prompt
- If either review finds issues → continue implementer session with
session_id
- Never skip either review stage
Three-Stage Planning Pipeline
For complex features, use the full planning pipeline:
- Metis (pre-analysis) → Detects ambiguities, classifies intent, identifies AI failure points
- Prometheus (planning) → Interviews user, explores codebase, builds detailed implementation plan
- Momus (plan review) → Verifies plan executability, checks references, validates completeness
Session Continuation
Always use session_id to continue a sub-agent's work:
// First dispatch → get session_id in result
task(category="quick", load_skills=["omo-test-driven-development"], description="Task 1", prompt="...", run_in_background=false)
// Result includes session_id: "ses_abc123"
// Continue same agent with full context preserved
task(session_id="ses_abc123", load_skills=["omo-test-driven-development"], description="Fix issues in Task 1", prompt="Fix: ...", run_in_background=false)
Background Dispatch for Parallel Work
Independent exploration/research tasks should run in background:
task(subagent_type="explore", description="Find auth patterns", prompt="...", run_in_background=true)
task(subagent_type="librarian", description="Research JWT best practices", prompt="...", run_in_background=true)
// Continue working → collect results later with background_output(task_id="...")
