testing-evidence-collector
Pass
Audited by Gen Agent Trust Hub on Apr 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions direct the agent to execute shell commands (ls, grep, cat) and a specific script (./qa-playwright-capture.sh) as part of its mandatory process. These commands operate on the local file system and internal network (localhost:8000).
- [EXTERNAL_DOWNLOADS]: The skill references and instructs the agent to execute an external script,
./qa-playwright-capture.sh, which is not provided within the skill's file set. Executing unverified local scripts is a potential security risk if the script's origin or contents are untrusted. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection. It ingests data from local source files (*.html, *.css, *.blade.php) and structured test results (test-results.json) without sanitization or boundary markers.
- Ingestion points: Reads contents from
public/qa-screenshots/test-results.jsonand performs recursive greps on project source files. - Boundary markers: Absent; the instructions do not specify delimiters to separate untrusted data from agent instructions.
- Capability inventory: The agent can execute shell commands and read/write to the local file system.
- Sanitization: Absent; the agent directly processes the output of file reads and grep operations.
Audit Metadata