cursor-thermo-nuclear-review
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and audit external code changes and PR discussions, which creates an inherent surface for indirect prompt injection.
- Ingestion points: Instructions in
SKILL.mddirect the agent to audit a "checked out branch," "changes in the diff," and external PR/MR discussions. - Boundary markers: There are no explicit delimiters or specific instructions to treat the ingested code or PR comments as untrusted or to disregard embedded instructions within that data.
- Capability inventory: The skill requires file system read access to audit the branch and utilizes the
ghandglabCLI tools to fetch discussion data. - Sanitization: The skill does not implement sanitization or validation logic for the external content it processes.
- [COMMAND_EXECUTION]: The skill instructions explicitly encourage the agent to execute shell commands to interact with version control platforms.
- Evidence:
SKILL.mdinstructs the agent to "check the PR/MR discussion using gh/glab cli" to gather context for its report.
Audit Metadata