This skill should be used when deciding whether to "delegate to codex", "hand off to codex", "use codex for this", or when Claude needs to determine if a task is appropriate for Codex vs handling directly. Provides decision criteria for task routing between Claude and Codex.
Provides decision criteria for routing tasks between Claude and Codex based on scope, urgency, and complexity.
npx claudepluginhub dkoosis/cc-pluginsThis skill inherits all available tools. When active, it can use any tool Claude has access to.
Decision framework for routing tasks between Claude Code and OpenAI's Codex agent.
Claude Code and Codex have complementary strengths. Claude excels at interactive, exploratory work requiring judgment. Codex excels at bounded, well-defined tasks that can run asynchronously. Effective delegation maximizes throughput by matching tasks to the right agent.
| Factor | Favor Codex | Favor Claude |
|---|---|---|
| Scope | Clearly bounded | Open-ended |
| Urgency | Can wait 5-30 min | Needed now |
| Iteration | One-shot execution | Back-and-forth likely |
| Complexity | Junior dev could do it | Requires judgment |
| Context | Self-contained | Needs conversation history |
| Output | Code/PR | Explanation/discussion |
QA Tasks
Clearly-Scoped Bug Fixes
Repetitive Multi-File Changes
Background Work
Architecture and Design
Exploratory Work
Interactive Tasks
Urgent Work
Is this blocking your current work?
├── Yes → Keep in Claude
└── No → Continue to Step 2
Can you write a clear CTM with:
- Specific success criteria
- Defined file scope
- Testable outcome
├── Yes → Continue to Step 3
└── No → Keep in Claude
Will this likely need back-and-forth?
├── Yes → Keep in Claude
└── No → Continue to Step 4
Could a competent junior dev do this with clear instructions?
├── Yes → Delegate to Codex
└── No → Keep in Claude
Claude should suggest Codex delegation when detecting:
When delegating to Codex:
/codex:submit: Hand off to Codex/codex:status to monitorUser: "Can you add test coverage for the new handler?"
→ Clear scope, QA task, can run async
→ Suggest: "This is a good Codex task. Generate CTM?"
User: "Fix all the golangci-lint errors"
→ Well-defined, repetitive, objectively complete
→ Suggest: "Perfect for Codex. Submit now?"
User: "Review the PR for security issues"
→ Checklist-based, bounded scope
→ Suggest: "Codex can handle this review"
User: "How should we structure the new API?"
→ Architecture decision, needs discussion
→ Handle directly, don't suggest Codex
User: "Debug why this test is flaky"
→ Investigation needed, unclear scope
→ Handle directly, explore interactively
User: "I need this fix in the next 5 minutes"
→ Urgent, can't wait for Codex
→ Handle directly, prioritize speed
Tasks delegated to Codex often map to:
After Codex completes, friction reports can be mined into Orca traps and rules for continuous improvement.
Creating algorithmic art using p5.js with seeded randomness and interactive parameter exploration. Use this when users request creating art using code, generative art, algorithmic art, flow fields, or particle systems. Create original algorithmic art rather than copying existing artists' work to avoid copyright violations.
Applies Anthropic's official brand colors and typography to any sort of artifact that may benefit from having Anthropic's look-and-feel. Use it when brand colors or style guidelines, visual formatting, or company design standards apply.
Create beautiful visual art in .png and .pdf documents using design philosophy. You should use this skill when the user asks to create a poster, piece of art, design, or other static piece. Create original visual designs, never copying existing artists' work to avoid copyright violations.