geo-fundamentals
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOWPROMPT_INJECTION
Full Analysis
- Indirect Prompt Injection (LOW): The geo_checker.py script analyzes content from project files and outputs summaries to the agent. An attacker could craft malicious file names or internal content to influence the agent's subsequent reasoning based on the audit results.
- Ingestion points: scripts/geo_checker.py reads files using Path.read_text().
- Boundary markers: Audit results are printed without clear delimiters or explicit instructions to ignore embedded instructions.
- Capability inventory: Script is limited to local file reading and regex pattern matching; no network or file-writing capabilities identified.
- Sanitization: Filenames and content matches are not sanitized before being printed to the output.
Audit Metadata