author-vitest-tests
Pass
Audited by Gen Agent Trust Hub on Jul 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The analysis subagent is instructed to run shell commands such as
grep -rq "<symbol>" <file>andgit log --oneline --grep="revert" -- <file>using symbol names extracted directly from pull request diffs or branch history. This poses a risk of command injection if the extracted symbols or file names contain shell metacharacters. - [PROMPT_INJECTION]: The skill processes untrusted external data (PR diffs and branch content) through an analysis subagent. This creates an attack surface for indirect prompt injection, where malicious instructions hidden in code comments or diff content could attempt to influence the agent's test plan generation or broader behavior.
- [DATA_EXFILTRATION]: The skill accesses project source code and metadata from GitHub (via
gh pr diff), but this is consistent with its stated purpose of authoring tests for the Positron repository and is performed within the local development context.
Audit Metadata