# Mode Selection

> For understanding what each mode does, which mode is best (most appropriate mode) for a given situation, and establishing a pattern for choosing when and how a mode delegates or passes operation to another mode

- Skill: `majiayu000/mode-selection` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds add majiayu000/mode-selection`
- Raw SKILL.md: https://api.skillmd.com/api/skills/majiayu000/mode-selection/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: majiayu000 (https://skillmd.com/u/majiayu000)
- Updated: 2026-09-09
- Page: https://skillmd.com/skills/majiayu000/mode-selection

---


# Mode selection

## Mode descriptions

**Location**
- Mode-specific rules live under `.roo/rules-*/0*-*.md`.

**For Planning & Orchestration**
- `/architect`: All-in-one planning. Create phases and tasks -> QA -> User Approval -> Switch to `/orchestrator`.
- `/planner-a`: Complex Planing Stage 1. Create phases -> Brainstorm -> Switch to `/planner-b`.
- `/planner-b`: Complex Planning Stage 2. Create detailed tasks -> User Approval -> Switch to `/planner-c`.
- `/planner-c`: Complex Planning Stage 3. QA -> Finalize -> User Approval -> Switch to `/orchestrator`.
- `/orchestrator`: Manage execution. Coordinate implementation modes. Log. Fullfill the plan.
**For Implementation & Ops**
- `/code`: Complex engineering, analysis, deep debugging.
- `/code-monkey`: Routine coding, strict instruction adherence.
- `/front-end`: UI implementation.
- `/tester`: Test creation and execution.
- `/debug`: Error investigation and diagnosis.
- `/githubber`: GitHub CLI operations.
- `/task-simple`: Small, isolated operations.
- `/ask`: General inquiries.

## Mode selection strategy
**Evaluate** the current `task`. If another mode is more appropriate, **pass** the `task` and payload to that mode.
**Payload Requirements**:
- **Context**: Include relevant bug/issue details. If a `plan` is active, include pertinent sections.
- **Instructions**: Provide specific, actionable implementation steps.
- **Completion Trigger**: Specify the exact command to return when the task is finished.
**Response Protocol**:
- Mandate a reply via the `result` parameter containing a concise summary of the outcome.

**Prioritize** budget-friendly modes in this order (Low to High):
1.  **Low Budget** (Renaming, moving files, simple text replacement, DB column copying)
    - Use `/task-simple`
2.  **Medium Budget** (Refactoring, simple function creation, writing)
    - Use `/code-monkey`
3.  **High Budget** (Complex modification, test creation and use, or if Medium fails)
    - Use `/code` or `/tester`
4.  **Highest Budget** (Debugging, or if High fails)
    - Use `/debug`
**Special Exception**:
- **Front-End Tasks** (Medium or High complexity): **Always use** `/front-end`