菜单
Explore an existing codebase, design a concrete feature architecture, implement in small phases, and verify with focused tests.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Build a threat model for a codebase — map assets, entry points, trust boundaries, and the threats that matter — and write THREAT_MODEL.md to focus later scanning and triage.
Behavior-preserving cleanup of recently changed code for reuse, clarity, consistency, and efficiency.
Verify, deduplicate, rank, and route raw security findings from VULN-FINDINGS.json, scanner output, or a markdown report, then write TRIAGE.json and TRIAGE.md.
Static source-code vulnerability scan that maps focus areas, fans out read-only security agents, and writes VULN-FINDINGS.json plus VULN-FINDINGS.md for triage.
High-precision review of code changes for real bugs, security regressions, and explicit project-guideline violations, with independent verification before reporting.
Generate minimal, behavior-preserving fixes for verified security findings, then verify each fix builds, passes tests, and closes the root cause.
| name | feature-dev |
| description | Explore an existing codebase, design a concrete feature architecture, implement in small phases, and verify with focused tests. |
| when-to-use | When the user asks to build a non-trivial feature or change and wants the agent to understand local patterns before editing. |
| arguments | ["request"] |
Deliver a feature by first understanding the existing system, then making a concrete architecture decision, then implementing and verifying in small slices. Use this for changes large enough that direct editing would risk missing project patterns.
${request} is the feature or change request. If it is empty, use the user's latest message.
Launch a code-explorer agent with a self-contained prompt:
AGENTS.md, CLAUDE.md) that apply;The explorer must cite file:line references and distinguish facts from inference.
Launch a code-architect agent with the request and explorer findings. Ask for one implementation blueprint, not a menu:
If the blueprint exposes ambiguity that changes behavior or scope, ask the user before editing. Otherwise proceed.
Work through the blueprint in small, reviewable phases:
Use test-engineer for tests when the behavior is broad, stateful, security-sensitive, or regression-prone. For a narrow change, add focused tests directly. When you delegate implementation or tests to a mutating subagent, give it a narrow file scope and tell it not to touch unrelated files.
After implementation:
code-reviewer agent on the diff for bugs, guideline violations, and security regressions.code-simplifier agent if the diff includes new abstractions, repeated logic, or changed hot paths.verification agent or directly if the commands are obvious.Summarize: