Strategic Planning with Prometheus

@plan "Implement rate limiting for all API endpoints"

---
generated: 2026-03-01
agent: prometheus
status: ready
---

# {Plan Title}

## Objective
Clear, measurable goal statement

## Background
Context: why this work, what led to it, current state

## Scope
### IN SCOPE
- Specific deliverable 1
- Specific deliverable 2

### OUT OF SCOPE
- Thing we explicitly won't do
- Future work deferred

## Requirements
### Functional
- Requirement 1 with acceptance criteria
- Requirement 2 with acceptance criteria

### Non-Functional
- Performance targets
- Security requirements
- Compatibility requirements

## Implementation Plan
### Phase 1: {Name}
**Goal:** What this phase achieves

**Tasks:**
1. **Task name** [`file/path.ts:123`]
   - Specific action
   - How to verify
   - References to existing patterns

### Phase 2: {Name}
...

## Testing Strategy
- Unit test coverage requirements
- Integration test scenarios
- Manual verification steps

## Verification Checklist
- [ ] All acceptance criteria met
- [ ] Tests pass
- [ ] No new lint/type errors
- [ ] Documentation updated

## Risks and Mitigations
| Risk | Impact | Mitigation |
|------|--------|------------|

## Reference Materials
- Link to relevant docs
- Existing implementation patterns
- External resources

Monday 9:00 AM
  → @plan "Build user authentication"
  → Prometheus creates plan
  → /start-work
  → Atlas starts execution, creates boulder.json
  → Task 1 complete, Task 2 in progress...
  → [Session crashes]

Monday 2:00 PM (NEW SESSION)
  → User opens new session
  → /start-work
  → [Hook reads boulder.json]
  → "Resuming 'Build user authentication' - 3 of 8 tasks complete"
  → Atlas continues from Task 3

{
  "sisyphus_agent": {
    "planner_enabled": true,      // Enable/disable Prometheus
    "replace_plan": true          // Demote default planner
  },
  
  "agents": {
    "prometheus": {
      "model": "anthropic/claude-opus-4-6",
      "thinking": {
        "type": "enabled",
        "budgetTokens": 32000       // Extended thinking for planning
      },
      "prompt_append": "Leverage deep & quick agents heavily, always in parallel."
    },
    
    "metis": {
      "model": "anthropic/claude-opus-4-6"
    },
    
    "momus": {
      "model": "openai/gpt-5.2"    // Different perspective from GPT
    }
  },
  
  "disabled_hooks": [
    // "prometheus-md-only"  // Remove write restrictions (not recommended)
  ]
}

@plan "Migrate from REST to GraphQL while maintaining REST compatibility"

Prometheus: Which endpoints should be migrated first? All at once or phased?
User: Phased. Start with user endpoints.

Prometheus: Should REST remain indefinitely or have a deprecation timeline?
User: Deprecation in 6 months, full removal in 12 months.

Prometheus: What GraphQL schema design patterns exist in the codebase?
User: None yet, this is the first GraphQL work.

Prometheus: [Launches explore agent to find similar projects]

@plan "Add real-time notifications with WebSockets"

Prometheus: What events should trigger notifications?
User: New messages, mentions, system alerts.

Prometheus: Should notifications persist when user is offline?
User: Yes, store in database and send when they reconnect.

Prometheus: Found existing Redis pub/sub setup. Use that for broadcasting?
User: Yes, perfect.

@plan "Design caching layer for high-traffic API"

Prometheus: Expected request volume?
User: 10k req/sec peak, 2k avg.

Prometheus: Read-heavy or write-heavy?
User: 90% reads, 10% writes.

Prometheus: Consistency requirements? Eventual consistency acceptable?
User: 5-second staleness acceptable for most data.

Prometheus: [Consults Oracle agent for architecture recommendations]

# Plan with Prometheus
@plan "Build admin dashboard with analytics"

# Execute with ultrawork intensity
/start-work
# Then type: ulw

@plan "Implement payment integration"
/ralph-loop "/start-work"