360 Secure SkillHub
360 安全技能中心

triage

该 Skill 按复现、诊断、验证和修复四步处理缺陷报告,可从参数或 GitHub issue 获取信息,并在不可复现、低置信度或属于预期行为时提前结束,最终返回结构化结果供编排器发布评论和管理标签。

withastro
热度 0.00
最近更新 2026-10-19 03:06:16
分类:开发依赖:代码执行 / 读写文件 / 第三方 API / 内部调用大模型应用场景:调试排错 / 代码生成 / 工作流

安装技能

复制下面的话术发给你的 Agent,它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill: npx secure-skills-cli add withastro/astro/triage
Metadata
name
triage
description
Triage a bug report. Reproduces the bug, diagnoses the root cause, verifies whether the behavior is intentional, and attempts a fix. Use when asked to "triage issue

Triage

Triage a bug report end-to-end: reproduce the bug, diagnose the root cause, verify whether the behavior is intentional, and attempt a fix.

General Rules

Do not get stuck on infrastructure problems. If a server won't start, a port is blocked, or a tool is unavailable in the CI environment — bail out after 2 attempts and write your report with the data you already have. Looping on server restarts or missing CLI tools wastes your entire time budget. A partial report with solid findings is infinitely more valuable than no report because you ran out of time fighting a stale process.

Input

You need either:

  • issueTitle and issueBody provided in args (preferred — use these directly as the bug report), OR
  • A GitHub issue number or URL mentioned in the conversation (use gh issue view to fetch details)

If a triageDir is provided in args, use that as the working directory for the triage. Otherwise, default to triage/gh-<issue_number> (if you have an issue number) or triage/current.

Step 1: Reproduce

Read and follow reproduce.md. Use a subagent for this step to isolate context.

After completing reproduction, check the result:

  • If the issue was skipped (host-specific, unsupported version, etc.) — skip to Output.
  • If the issue was not reproducible — skip to Output.
  • If the issue was reproduced — continue to Step 2.

Step 2: Diagnose

Read and follow diagnose.md. Use a subagent for this step to isolate context.

After completing diagnosis, check your confidence:

  • If confidence is low — skip to Output.
  • If confidence is medium or high — continue to Step 3.

Step 3: Verify

Read and follow verify.md. Use a subagent for this step to isolate context.

After completing verification, check the verdict:

  • If the verdict is intended-behavior — skip to Output. The issue is not a bug; do not attempt a fix.
  • If the verdict is bug or unclear — continue to Step 4.

Step 4: Fix

Read and follow fix.md. Use a subagent for this step to isolate context.

Whether the fix succeeds or fails, continue to Output.

Output

After completing the triage (or exiting early), return your structured results so the orchestrator can post a comment and manage labels.