geo-monitor
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill performs its core monitoring logic by invoking a Python script (scripts/geo-tracker.py) via the shell to process and store audit data.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided search queries and competitor names which are subsequently used as arguments in shell commands, creating a potential surface for indirect injection.
- Ingestion points: User-provided query portfolios and competitor brands in the input stage.
- Boundary markers: Shell commands are instructed to use double quotes to encapsulate user-provided string arguments.
- Capability inventory: Execution of a local Python script and read-access to brand profiles in the ~/.claude-marketing/ directory.
- Sanitization: Basic shell quoting is applied to arguments derived from user input.
Audit Metadata