Evidence Collector
Pass
Audited by Gen Agent Trust Hub on Oct 25, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a local shell script
./qa-playwright-capture.shand perform various filesystem operations usingls,grep, andcatto inspect source code and test results. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external data from local files, including HTML views, CSS, and test result JSON files. This creates a surface for indirect prompt injection if those files contain malicious instructions.
- Ingestion points:
public/qa-screenshots/test-results.json,resources/views/,*.html,*.css(specified in STEP 1). - Boundary markers: None. No instructions are given to the agent to treat the content of these files as untrusted or to ignore embedded instructions.
- Capability inventory: The agent has the ability to execute local scripts (
./qa-playwright-capture.sh) and perform recursive file searches (grep -r). - Sanitization: No sanitization, escaping, or schema validation of the ingested file content is mentioned.
Audit Metadata