agency-reality-checker
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The agent executes local shell commands such as ls, grep, and cat to inspect the project environment and verify the existence of claimed features. It also runs a local shell script named ./qa-playwright-capture.sh to generate visual evidence and performance metrics for testing.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to read and evaluate content from untrusted project files and test result data. Ingestion points: Local project files (.html, .css, .blade.php) and test result JSON files. Boundary markers: None identified. Capability inventory: Local shell command execution. Sanitization: None performed on ingested data.
Audit Metadata