linkfox-zhihuiya-description-data

Fail

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: HIGHPERSISTENCEINDIRECT_PROMPT_INJECTIONCREDENTIALS_UNSAFEOBFUSCATIONCOMMAND_EXECUTIONDYNAMIC_EXECUTION
Full Analysis
  • [PERSISTENCE]: The file references/onboarding.md provides explicit instructions for the agent to guide users in persisting API keys by appending export commands to shell configuration profiles, including ~/.bashrc, ~/.zshrc, and ~/.bash_profile. This approach to credential management constitutes a persistence mechanism.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes detailed patent descriptions and specifications retrieved from the Zhihuiya database, which are external technical documents that can influence agent behavior.
  • Ingestion points: The description field fetched by scripts/zhihuiya_description_data.py from the /zhihuiya/descriptionData endpoint.
  • Boundary markers: None; the technical content is rendered directly to stdout for the agent to process.
  • Capability inventory: The skill environment allows for file system writes to the project directory and network communication with vendor-provided APIs.
  • Sanitization: There is no evidence of content validation or sanitization of the retrieved patent text before it is presented to the agent's context.
  • [DYNAMIC_EXECUTION]: Both scripts/onboarding.py and scripts/zhihuiya_description_data.py use sys.path.insert(0, ...) to dynamically modify the module search path to include directories outside the immediate skill folder (e.g., ../../_shared) for importing shared utilities.
  • [OBFUSCATION]: The script scripts/onboarding.py contains a hardcoded Base64 string eyJhX2lkIjoiNmEyMmM4YjA1YmM5MTZhIiwiZF9pZCI6IiJ9 (decodes to a JSON object {"a_id":"6a22c8b05bc916a","d_id":""}) used as a static identifier in authentication headers.
  • [CREDENTIALS_UNSAFE]: The skill is designed to manage sensitive credentials (LINKFOX_AGENT_API_KEY and LINKFOXAGENT_API_KEY) and includes a login flow script (scripts/onboarding.py) that handles session tokens and API keys.
  • [COMMAND_EXECUTION]: The onboarding documentation specifies the use of shell commands such as setx and redirected echo commands to configure environment variables in the user's system.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Sep 15, 2026, 07:12 AM