linkfox-zhihuiya-claim-data-translated
Pass
Audited by Gen Agent Trust Hub on Aug 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [PERSISTENCE_MECHANISMS]: The onboarding instructions in
references/onboarding.mdand logic inscripts/onboarding.pyguide the user or agent to execute shell commands that modify persistent system configuration files. Specifically, it suggests appending environment variable exports to~/.zshrcand~/.bashrc, and usingsetxon Windows to persist theLINKFOX_AGENT_API_KEY. While these actions are intended for legitimate credential management, they constitute persistence mechanisms by modifying shell initialization scripts. - [INDIRECT_PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it ingests and processes untrusted patent claim data from an external API.
- Ingestion points: The
call_apifunction inscripts/zhihuiya_claim_translated.pyretrieves patent claim text from the Zhihuiya database, which is external to the agent's immediate control. - Boundary markers: The skill does not implement explicit boundary markers or instructions to the agent to ignore potentially malicious instructions embedded within the retrieved patent claims.
- Capability inventory: The skill environment includes capabilities for network requests and shell command execution via the onboarding scripts.
- Sanitization: There is no evidence of sanitization or filtering of the retrieved patent claim text before it is interpolated into the agent's context.
Audit Metadata