kimi-find-skills
Pass
Audited by Gen Agent Trust Hub on Nov 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONREMOTE_CODE_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes content from external web sources which could contain malicious instructions designed to influence the agent's behavior.
- Ingestion points: Data is retrieved via
web_search(GitHub) andbrowser_visit(https://skillhub.cn/) to evaluate skill relevance and descriptions. - Boundary markers: The skill lacks explicit boundary markers or delimiters to isolate untrusted web content from its core logic, increasing the risk that the agent may follow instructions embedded within external search results.
- Capability inventory: The agent possesses capabilities including
shellcommand execution and web browsing, which could be abused if an indirect injection is successful. - Sanitization: No sanitization or validation mechanisms are described for the content retrieved from external sites before it is processed by the agent.
- [REMOTE_CODE_EXECUTION]: The skill facilitates the discovery of unvetted third-party code (agent skills) from external repositories.
- Source discovery: It identifies skills on GitHub and SkillHub based on user-controlled keywords and metadata like star counts, which can be manipulated by malicious actors to promote harmful content.
- Recommendation risk: By providing links and installation instructions for third-party skills, the agent may lead users to install executable logic or prompts from unverified sources, potentially compromising the host environment.
Audit Metadata