AgentSkillsCN

test

运行测试，分析失败原因，并提出修复方案

SKILL.md
--- frontmatter
name: test
description: Run tests, analyze failures, and propose fixes
argument-hint: "[unit|integration|agent|all|<file path>]"
user-invocable: true
allowed-tools: "Read, Grep, Glob, Bash(make test*), Bash(make lint*), Bash(make check*)"
context: fork
agent: general-purpose

Run the test suite, analyze results, and propose fixes for any failures.

Scope

  • •$ARGUMENTS is empty or "all": run make test (full suite, fail-fast by tier).
  • •$ARGUMENTS is "unit": run make test-unit.
  • •$ARGUMENTS is "integration": run make test-integration.
  • •$ARGUMENTS is "agent": run make test-agent.
  • •$ARGUMENTS is a file path: run make test-file FILE=<path>.

Process

  1. •Run the specified tests.
  2. •If all pass: report success with a brief summary.
  3. •If any fail:
    • •Parse the error output to identify the root cause.
    • •Read the failing test and the code under test.
    • •Propose a fix — explain what went wrong and how to resolve it.
    • •Do NOT apply the fix automatically — present it for approval.

Output

Write detailed results to scratch/test_results_latest.md with:

  • •Pass/fail counts per tier
  • •Failure details (test name, error message, relevant code)
  • •Proposed fixes if applicable

Return a ≤5 line summary to the main context.