aibtc-news-protocol
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted free-form text input ('observations') to generate news signals and shell commands. This creates a surface where malicious input could attempt to manipulate the generated signal content.\n
- Ingestion points:
aibtc-news-protocol.tsreceives untrusted data through the--observationcommand-line argument.\n - Boundary markers: No explicit delimiters are used to wrap the untrusted observation in the output content.\n
- Capability inventory: The skill can perform network HEAD requests and generates shell command strings intended for execution via another skill.\n
- Sanitization: The implementation uses single-quote escaping when building command strings to mitigate basic command injection attempts.\n- [EXTERNAL_DOWNLOADS]: The
check-sourcessubcommand allows the agent to perform network HEAD requests to arbitrary URLs provided in the arguments. This is a functional requirement for verifying news sources but could be used to probe network resources.\n- [COMMAND_EXECUTION]: The skill generates shell command strings for theaibtc-newsCLI based on user-provided input. While sanitized, the generation of executable command strings from external data is a pattern that requires caution in autonomous agents.
Audit Metadata