cursor-principle-never-block-on-the-human
Pass
Audited by Gen Agent Trust Hub on Oct 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The instructions encourage the agent to bypass human confirmation for "reversible" work. While aimed at efficiency, this behavioral pattern creates a surface for indirect prompt injection where the agent might execute instructions found in untrusted data (like code comments or documentation) before a human can intervene.\n- Ingestion points: The agent is expected to process and modify files such as source code, markdown notes, and task descriptions.\n- Boundary markers: The skill explicitly lists high-risk, irreversible actions (force-push, data deletion, external messaging) that still require human confirmation, which serves as a safety constraint.\n- Capability inventory: The skill assumes capabilities for writing/editing files and managing project tasks.\n- Sanitization: There are no instructions provided for sanitizing or validating the content of files before acting upon them.
Audit Metadata