盘古·skill

PassAudited by ClawScan on May 3, 2026.

Overview

This instruction-only skill looks coherent and non-malicious, but it can read and distill your knowledge-base content into reusable skill files, so review the scope and output carefully.

This skill appears safe to install as an instruction-only workflow, but use it with intentional scope: choose a specific knowledge base, folder, or document; remove confidential, personal, or copyrighted material you should not share; and review all generated files before zipping, uploading, or installing the resulting Skill.

Findings (3)

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.

What this means

Private notes, proprietary documents, or prior-chat details could be summarized into generated Skill files that may later be reused, zipped, uploaded, or shared.

Why it was flagged

The skill may process an entire knowledge base or prior conversation and turn that material into persistent Skill content; this is central to its purpose but can carry private or prompt-like content into future reuse.

Skill content
读取策略:- 完整蒸馏:读取知识库所有文件和文件夹结构 ... - 任务蒸馏:分析对话历史,提取完整工作流
Recommendation

Use a clearly bounded source, exclude secrets and personal data, and review the generated SKILL.md, references, assets, and examples before installing or sharing them.

What this means

The agent may create or modify files in the chosen Skill directory as part of normal use.

Why it was flagged

The instructions direct the agent to create a Skill directory and write files after confirmation. That is purpose-aligned, but it is still local/project file mutation.

Skill content
Phase 1.5: 创建 Skill 目录 ... 收到确认后立即执行 ... 所有参考文件必须存在 skill 目录内部
Recommendation

Confirm the output path, keep backups for existing Skills, and inspect file diffs before registering or installing the generated Skill.

What this means

Any referenced knowledge base that the agent can access may be used as source material for the generated Skill.

Why it was flagged

The skill relies on the user agent's existing access to an IMA knowledge base. No credentials or tokens are requested, and this access is expected for the stated purpose.

Skill content
本技能自动读取 IMA 知识库内容并蒸馏 ... 知识库必须可访问 — 使用 `@知识库名称` 引用
Recommendation

Reference only knowledge bases you own or are authorized to process, and avoid granting broader workspace access than needed.