AgentSkillsCN

task-router

根据任务类型，自动将任务分派给相应的专业子Agent。负责测试执行、脚本运行，以及代码分析的委派工作。

SKILL.md
--- frontmatter
name: task-router
description: Automatically routes tasks to specialized subagents based on task type. Handles test execution, script running, and code analysis delegation.
user-invocable: false

Task Router Skill

This skill automatically delegates tasks to subagents when Claude detects specific patterns in user requests.

Delegation Rules

Script Runner Subagent

Delegate to Script Runner when the task involves:

  • •Running Python scripts in the scripts/ directory
  • •Database migrations or seeding operations
  • •Data cleanup or maintenance tasks
  • •Performance profiling or analysis scripts
  • •Cost analysis or reporting scripts
  • •Any command starting with python scripts/

Delegation pattern:

code
/script-runner [describe what script needs to run and why]

Test Runner Subagent

Delegate to Test Runner when the task involves:

  • •Running pytest or test suites
  • •Verifying bug fixes (e.g., "verify BUG-001 is fixed")
  • •Checking test coverage
  • •Running specific test files or test functions
  • •Any command containing pytest

Delegation pattern:

code
/test-runner [describe what tests to run and what to verify]

Detection Patterns

Script execution triggers:

  • •"run the [script name] script"
  • •"execute the migration"
  • •"seed the database"
  • •"clean up [data type]"
  • •"analyze performance"
  • •"generate a report"

Test execution triggers:

  • •"run tests"
  • •"verify [bug/feature]"
  • •"check test coverage"
  • •"test the [component]"
  • •"make sure tests pass"

When NOT to delegate

Do NOT delegate for:

  • •Code review or analysis (handle inline)
  • •Writing new code (handle inline)
  • •Explaining how code works (handle inline)
  • •Simple file operations (handle inline)
  • •Questions about the codebase (handle inline)

Example Delegations

User: "Run the narrative fingerprinting script to regenerate all fingerprints" Action: /script-runner Run scripts/regenerate_fingerprints.py to update all narrative fingerprints in the database

User: "Make sure the timeline tests pass after my changes" Action: /test-runner Run pytest tests/test_timeline.py to verify timeline functionality still works

User: "Verify BUG-001 is actually fixed" Action: /test-runner Run the full test suite and specifically verify article count matching works correctly

Context to Provide

When delegating, always include:

  1. •What specific action needs to happen
  2. •What files/components are involved
  3. •What success looks like
  4. •Any relevant context from the current conversation

Integration with Sprint Workflow

When tasks relate to sprint items (e.g., BUG-001, BUG-002, FEATURE-003):

  • •Include the sprint item ID in the delegation
  • •Mention what needs to be verified
  • •Reference any acceptance criteria