skills-security-check
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONMETADATA_POISONING
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is specifically designed to ingest and process untrusted external data (other skills, scripts, and documentation). This creates an attack surface where a malicious skill being audited could attempt to inject instructions into the auditor agent. The skill includes explicit boundary markers, instructions to treat input as static text, and a whitelist of read-only tools to mitigate this risk.
- [METADATA_POISONING]: There is a potential inconsistency in the skill metadata; the description claims the tool is produced by 'Tencent YunDing Lab' (腾讯云鼎实验室), but the registered author name is 'tugoukezhang'. While not inherently malicious, this attribution mismatch is noted as a metadata inconsistency.
- [REMOTE_CODE_EXECUTION]: The skill mentions and provides instructions for analyzing remote scripts using
web_fetch. However, these instructions are explicitly for static analysis purposes (reading text only) and do not involve execution of the remote content. - [COMMAND_EXECUTION]: The skill contains numerous dangerous command patterns (e.g.,
curl | bash,rm -rf /,sudo). These are included exclusively as diagnostic patterns for the auditor to detect in other files and are not instructions for the agent to execute them itself.
Audit Metadata