PsyVector pv52

AdvisoryAudited by Static analysis on Apr 13, 2026.

Overview

No suspicious patterns detected.

Findings (0)

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

The agent may send crisis-related queries to search providers and fetch external pages as part of answering.

Why it was flagged

The skill explicitly requires the agent to use external research tools during crisis analysis.

Skill content
Step 2: 研究(必须使用工具)... 调用web_search搜索同类危机处理案例 + web_fetch获取事件最新进展
Recommendation

Use web tools only when current research is needed, and avoid including confidential details in search queries.

ConcernMedium Confidence
ASI06: Memory and Context Poisoning
What this means

If the agent has access to company or private data tools, it could pull broader internal information into the conversation than the user intended.

Why it was flagged

For chronic crises, the skill tells the agent to retrieve internal data if available, but does not define which internal sources are allowed, what approval is required, or what data should be excluded.

Skill content
慢性型:调用web_search搜索行业案例 + 内部数据调取(如可用)
Recommendation

Restrict connected data tools, require explicit user confirmation before internal data retrieval, and define allowed sources and sensitive-data exclusions.