search-x
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill accesses a local configuration file to retrieve necessary API credentials for its operation.
- Evidence: The script
scripts/search.jsreads from~/.clawdbot/clawdbot.jsonto retrieve thexAIAPI key. - Context: Accessing the platform's configuration file for stored skill settings is standard behavior; credentials are only transmitted to the verified vendor service endpoint (
api.x.ai). - [INDIRECT_PROMPT_INJECTION]: The skill acts as a proxy for external social media content, which introduces an indirect prompt injection surface when the agent processes search results.
- Ingestion points: Tweet data is fetched from the xAI API and processed within
scripts/search.js. - Boundary markers: None; external content is printed directly to the output stream without specific delimiters or warnings for the agent.
- Capability inventory: The skill uses filesystem access for configuration reading and HTTPS for network communication with the API provider.
- Sanitization: External tweet content is displayed without sanitization, relying on the underlying model's inherent safety filters and guidelines.
Audit Metadata