메뉴
Choose the right subagent for a task and write a clear delegation prompt including issue context and success criteria.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
| name | subagent-delegation |
| description | Choose the right subagent for a task and write a clear delegation prompt including issue context and success criteria. |
This skill defines when to use which subagent and how to write the delegation prompt so the subagent has full context and does not need to open the issue again.
| Task type / keywords | Subagent |
|---|---|
| Bugs, errors, panics, reproduction, stack traces | debugger |
| Run/fix tests, clippy, CI, test failures | test-runner |
| Code review, safety, style, best practices | code-reviewer |
| Benchmarks, performance analysis, throughput | benchmarker |
| Docs, examples, API comments, readme | documenter |
| UI/UX, accessibility, motion, layout, design | ui-ux-reviewer |
| Verify completed work, re-run checks, validate | verifier |
| Release, versioning, changelog, semver | cargo-release |
When the issue spans multiple areas (e.g. fix a bug and add tests), delegate in sequence: e.g. debugger then test-runner, then optionally verifier. State the order in your summary to the user.
The prompt you pass to the subagent must include:
foo.rs and ensure cargo test passes."Rule: the subagent should be able to do the job without opening the issue again.
Use the platform's subagent/task invocation tool (e.g. mcp_task). Pass:
rust-debugger, test-runner, code-reviewer, benchmarker, documenter, ui-ux-reviewer, verifier, cargo-release). Use the exact subagent type names supported by your environment.Do not invent tool or parameter names; use the names provided by your environment for the subagent/task tool.
Validates animation durations, enforces typography scale, checks component accessibility, and prevents layout anti-patterns in Tailwind CSS projects. Use when building UI components, reviewing CSS utilities, or enforcing design consistency.
Making changes to the app's functionality.
Create Criterion benchmarks with proper group macros and configuration. Use when writing benchmarks, creating benchmark groups, using criterion_group! and criterion_main! macros, setting throughput, or benchmarking multiple implementations. Handles benchmark groups, throughput metrics, conditional compilation, and proper black_box usage.
Read and parse GitHub issues; extract title, body, labels, checklists, and acceptance criteria to support triage and delegation.
Apply cognitive and perception-based UI principles from Laws of UX. Use when evaluating layout, interaction design, target size and distance, choice count, information grouping, memory load, or response time. Cites specific laws (e.g. Fitts's Law, Hick's Law) to justify design decisions or suggest improvements.
Work with async/await, futures, and asynchronous Rust code. Use when writing async functions, working with tokio or async-std, understanding futures, or implementing async traits. Handles async/await syntax, futures, executors, pinning, Send/Sync bounds, and async patterns.