geo-prospect
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input from users that could potentially influence agent behavior if malicious content is stored and later retrieved.
- Ingestion points: Prospect domains, contact names, interaction notes, and reasons for lost deals are ingested via user commands and stored in '~/.geo-prospects/prospects.json' (SKILL.md).
- Boundary markers: None are employed; data is appended directly to the JSON structure without delimiters or instructions to ignore embedded commands.
- Capability inventory: The skill utilizes 'Bash', 'Write', and 'Read' tools, providing a broad surface for system interaction (SKILL.md).
- Sanitization: There is no definition of validation or sanitization for input strings before they are saved to persistent storage.
- [COMMAND_EXECUTION]: The skill leverages the 'Bash' tool to perform file system operations and execute external utility commands.
- The '/geo prospect audit' command invokes an external binary '/geo quick', creating a functional dependency on the host environment.
- It performs shell operations such as 'mkdir -p' to establish its persistent data storage directory.
Audit Metadata