Menu
Coordinates built-in and custom agents when work can be parallelized or staged. Use when one agent is not enough for the task.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Automatic quality control, linting, and static analysis procedures. Use when validating code changes for syntax correctness and project standards. Triggers on keywords: lint, format, check, validate, types, static analysis.
Refactoring and cleanup - improving code structure, removing dead code, eliminating duplication, or pre-merge cleanup. Use when code is hard to maintain, has dead code or debug artifacts, or needs pre-merge polishing.
Fix a GitHub issue end-to-end - from reading the issue, understanding requirements, implementing the fix, writing tests, and creating a commit. Use when given a GitHub issue number or URL to resolve.
Optimize code for readability, performance, maintainability, and security across Bash, Python, and Rust. Use when asked to improve code quality, optimize performance, add type safety, or refactor for idioms.
Create, debug, and optimize GitHub Actions workflows with security best practices. Use when asked to "create workflow", "fix workflow", "add CI", or needs help with GitHub Actions.
Review code changes (diffs, PRs, patches) and provide structured, actionable feedback on correctness, maintainability, and test coverage. Use when the user asks for a code review, requests feedback on a patch/PR, or wants an assessment of changes.
| name | parallel-agents |
| description | Coordinates built-in and custom agents when work can be parallelized or staged. Use when one agent is not enough for the task. |
Use this skill to split independent workstreams, route tasks to the right agent, and synthesize one final answer. Keep the plan grounded in the agents that actually exist in this repository.
Use this skill when the task naturally separates into discovery, research, implementation, or review streams.
| Good fit | Avoid |
|---|---|
| Exploration plus implementation plus review | Small single-file edits |
| Independent subproblems that can run in parallel | Tasks where one agent already has the full context |
| Domain handoffs: research -> code -> review | Busywork that only adds coordination overhead |
| Need | Agent |
|---|---|
| Broad repo discovery | explore or custom explorer |
| Architecture and task breakdown | planner |
| External best-practice research | researcher |
| Implementation | coder |
| Critical review | reviewer |
| Bug investigation | debug |
| UI work | frontend-specialist |
| Workflow/CI work | workflow-engineer |
| AI config or repo structure | repo-architect |
| Cleanup and dead code removal | codebase-maintainer or janitor |
| Git workflow help | git |
| Pattern | Sequence |
|---|---|
| Full feature | explorer -> planner -> researcher -> coder -> reviewer |
| Focused bug fix | debug -> coder/reviewer |
| Repo cleanup | codebase-maintainer -> reviewer |
| Workflow change | workflow-engineer -> reviewer |
explorer maps affected files.planner breaks work into waves.coder implements the approved plan.reviewer audits the result.researcher investigates an external library.workflow-engineer checks CI implications.