app-review-mining
Pass
Audited by Gen Agent Trust Hub on Oct 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses a bash script (scripts/crawlora.sh) to perform API requests via curl. The script is well-hardened: it restricts methods to GET, validates API key format, and enforces a whitelist of allowed API routes. It also ensures the API key is not visible in process lists by using a temporary configuration file with restricted permissions.
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from external sources (App Store and Google Play) which contains user-generated content like reviews and descriptions. This presents an indirect prompt injection surface. 1. Ingestion points: JSON output from the Crawlora API fetched via scripts/crawlora.sh. 2. Boundary markers: No explicit markers are defined in the instructions for output processing. 3. Capability inventory: The skill has network and shell execution capabilities via the helper script. 4. Sanitization: The helper script performs URL-encoding of all input parameters and rejects potentially dangerous characters like the @ symbol.
Audit Metadata