Skip to main content

​
/multi-plan Command

Multi-model collaborative planning - Context retrieval + Dual-model analysis (Codex + Gemini) → Generate step-by-step implementation plan.

​
Core Protocols

  • Language Protocol: Use English when interacting with tools/models
  • Mandatory Parallel: Codex/Gemini calls MUST use run_in_background: true
  • Code Sovereignty: External models have zero filesystem write access
  • Planning Only: This command allows reading context and writing plan files, but NEVER modify production code

​
Command Syntax

/multi-plan <feature description>
​
feature description
string
required
Description of what needs to be planned

​
Execution Workflow

​
Phase 1: Full Context Retrieval

  1. Prompt Enhancement - Call mcp__ace-tool__enhance_prompt
  2. Context Retrieval - Call mcp__ace-tool__search_context
  3. Completeness Check - Ensure complete definitions
  4. Requirement Alignment - Clarify ambiguities

​
Phase 2: Multi-Model Collaborative Analysis

  1. Distribute Inputs - Parallel call Codex and Gemini
    • Codex: Technical feasibility, architecture, risks
    • Gemini: UI/UX impact, visual design
  2. Cross-Validation - Integrate perspectives
  3. Dual-Model Plan Draft (Optional) - Both models output plan drafts
  4. Generate Implementation Plan - Claude synthesizes final plan

​
Model Roles

ModelAuthorityFocus
CodexBackendTechnical feasibility, architecture impact, performance
GeminiFrontendUI/UX impact, user experience, visual design
ClaudeOrchestratorSynthesize plans, coordinate workflow

​
Plan Output

## Implementation Plan: <Task Name>

### Task Type
- [ ] Frontend (→ Gemini)
- [x] Backend (→ Codex)
- [ ] Fullstack (→ Parallel)

### Technical Solution
<Optimal solution synthesized from Codex + Gemini analysis>

### Implementation Steps
1. <Step 1> - Expected deliverable
2. <Step 2> - Expected deliverable
...

### Key Files
| File | Operation | Description |
|------|-----------|-------------|
| path/to/file.ts:L10-L50 | Modify | Description |

### Risks and Mitigation
| Risk | Mitigation |
|------|------------|

### SESSION_ID (for /multi-execute use)
- CODEX_SESSION: <session_id>
- GEMINI_SESSION: <session_id>

​
Plan Delivery

After planning:
  1. Present complete implementation plan
  2. Save plan to .claude/plan/<feature-name>.md
  3. Output prompt:
**Plan generated and saved to `.claude/plan/feature-name.md`**

**Please review the plan above. You can:**
- **Modify plan**: Tell me what needs adjustment
- **Execute plan**: Copy the following command to a new session

```bash
/multi-execute .claude/plan/feature-name.md


4. Immediately terminate (no auto-execution)

## Absolutely Forbidden

- Ask user "Y/N" then auto-execute
- Any write operations to production code
- Automatically call `/multi-execute`
- Continue triggering model calls without explicit request

## Examples

```bash
/multi-plan Add real-time notifications when markets resolve
  • Commands: /multi-execute, /multi-workflow, /plan