geo-technical
Pass
Audited by Gen Agent Trust Hub on Aug 9, 2026
Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill instructions direct the agent to use
curlandWebFetchto perform network operations on arbitrary target URLs provided by the user. These requests target domains outside of the standard allow-list to retrieve HTML content and HTTP headers for auditing purposes. - [PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection because it ingests and processes data from external websites.
- Ingestion points: Raw HTML and HTTP headers are fetched from external websites and incorporated into the agent's context for analysis (SKILL.md).
- Boundary markers: The instructions do not specify the use of delimiters or 'ignore embedded instructions' prompts to isolate the audited content from the agent's logic.
- Capability inventory: The agent has the capability to perform network requests (
curl) and write analysis results to the filesystem (GEO-TECHNICAL-AUDIT.md). - Sanitization: The skill lacks instructions for sanitizing, escaping, or filtering the retrieved external content before it is processed by the model.
Audit Metadata