From ecc
Use this skill to determine the correct sub-agent type for task delegation. Provides routing tables for implementation tasks and review comment fixes, along with standard invocation patterns and error handling.
How this skill is triggered — by the user, by Claude, or both
Slash command
/ecc:agent-routerThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
This skill defines routing tables for selecting the appropriate sub-agent when delegating tasks or review comment fixes.
This skill defines routing tables for selecting the appropriate sub-agent when delegating tasks or review comment fixes.
subagent_type for a Task tool callImplementation tasks should be routed based on the nature of the work. Default to general-purpose unless a clearly better match exists.
| Task Nature | subagent_type | Selection Criteria |
|---|---|---|
| General feature implementation | general-purpose | Default. Composite or mixed tasks |
| Test implementation focused | ecc:tdd-guide | Tasks centered on adding/modifying tests |
| E2E testing | ecc:e2e-runner | Playwright E2E tests |
| Build/type error fixing | ecc:build-error-resolver | Build errors or type errors |
| Refactoring/dead code removal | ecc:refactor-cleaner | Code cleanup, removing unused code |
| Security tasks | ecc:security-reviewer | Auth, input validation, vulnerability fixes |
| Kotlin/Quarkus implementation | ecc:kotlin-coder | Kotlin code, Quarkus config, native image |
| SQL/schema design | ecc:sql-coder | SQL, migrations, query optimization |
| JPA/Hibernate entities | ecc:jpa-model-coder | Entity design, relations, N+1 prevention |
| TypeScript/Next.js implementation | ecc:nextjs-coder | React/Next.js components, Server Actions, API routes |
| Terraform/IaC implementation | ecc:terraform-coder | Terraform config, modules, cloud infra, state management |
| Documentation updates | ecc:doc-updater | Documentation and codemap sync |
| Architecture decisions | ecc:architect | System design, architectural review |
Review comment fixes should be routed based on the category assigned by the review comment parser.
| Comment Category | subagent_type | Selection Criteria |
|---|---|---|
| security | ecc:security-reviewer | Security vulnerability fixes |
| performance | Domain-specific agent (*) | Performance improvements |
| bug | general-purpose | Bug fixes |
| code-quality | ecc:refactor-cleaner | Code quality improvements |
| testing | ecc:tdd-guide | Test additions/modifications |
| style | general-purpose | Style and naming fixes |
| (other) | general-purpose | Any other category |
(*) For performance comments, cross-reference the Task-Type Routing Table to select the domain-specific agent. For example: SQL performance → ecc:sql-coder, Kotlin performance → ecc:kotlin-coder. If no domain match, fall back to general-purpose.
Use the Task tool with the following structure:
Task tool:
- prompt: <task description + relevant context>
- subagent_type: <selected from routing table>
- description: <brief description for tracking>
Implementation tasks — include in prompt:
Review comment fixes — include in prompt:
| Situation | Action |
|---|---|
| Sub-agent fix fails | Report error, mark as unresolved, continue with next item |
| Sub-agent times out | Retry once with simplified prompt; if still fails, mark as unresolved |
| Wrong agent selected (output doesn't match expectations) | Re-delegate to general-purpose as fallback |
| Multiple related issues | Group into a single sub-agent call for coherent fixes |
npx claudepluginhub nisioka/everything-claude-codeCreates bite-sized, testable implementation plans from specs or requirements, with file structure and task decomposition. Activates before coding multi-step tasks.