Spec Coding

PassAudited by ClawScan on May 1, 2026.

Overview

This is a benign instruction-only skill for drafting software specification documents, with only expected cautions around reading requested code and optional sub-agent handoffs.

This skill appears safe for drafting and reviewing specs. Before using it on an existing project, scope the files you want analyzed and confirm any sub-agent handoff or implementation step, especially for proprietary code.

Findings (2)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

Info

#ASI02: Tool Misuse and Exploitation

What this means

The assistant may inspect project code when asked to generate specs for an existing codebase.

Why it was flagged

The skill may read existing project source code to infer specifications. This is expected for the stated reverse-specification use case, but users should ensure only intended files are provided.

Skill content

老项目逆向生成规格... 1. **读取源代码** - 功能代码 → 分析输入/输出/行为

Recommendation

Use the skill on the intended repository or files, and avoid including secrets or unrelated private code in the context.

Low

#ASI07: Insecure Inter-Agent Communication

What this means

If sub-agent features are enabled, project details could be passed to other agent roles or an implementation agent could make code changes after user approval.

Why it was flagged

The skill describes optional collaboration with sub-agents, including research, copyediting, and implementation. It does not specify a concrete protocol or automatic handoff, but users should be aware that specs or code context could be shared with other agents if their environment supports that.

Skill content

## 与子代理协作
- **研究员**: 搜集竞品规格、行业标准
- **文案师**: 润色用户提示文案、错误消息
- **落地执行**: 根据规格实现代码、编写测试

Recommendation

Confirm any sub-agent handoff before proceeding, especially before sharing proprietary code or allowing implementation changes.