unity-test-runner
Pass
Audited by Gen Agent Trust Hub on Feb 15, 2026
Risk Level: LOWPROMPT_INJECTION
Full Analysis
- [Indirect Prompt Injection] (LOW): The script
scripts/parse-test-results.jsis designed to ingest and parse NUnit XML data from external test runs, creating an attack surface where malicious instructions could be injected into the agent's reasoning loop. - Ingestion points: The script reads raw XML content from a file path provided via command-line arguments using
fs.readFileSync(line 287). - Boundary markers: Absent. The parsed results (test names, failure messages, and stack traces) are printed directly to the console, where they are likely to be consumed by an AI agent without clear delimiters.
- Capability inventory: No subprocess execution, file-writing, or network operations were found in the provided scripts. The risk is limited to influencing the agent's internal state or subsequent decisions.
- Sanitization: Absent. Content extracted from
<test-case>,<message>, and<stack-trace>tags is passed to the output using regex matching without filtering or escaping potentially malicious natural language instructions. - [Data Exposure] (INFO):
scripts/find-unity-editor.jsaccesses environment variables (HOME,USERPROFILE) and standard system paths to locate Unity installations. This is limited to local path discovery necessary for the skill's stated purpose and does not involve data exfiltration.
Audit Metadata