principle-never-block-on-the-human
Pass
Audited by Gen Agent Trust Hub on Nov 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's core principle is to "never block on the human," directing the agent to proceed with execution and present results later rather than asking for permission. This instruction to stay unblocked directly reduces the human-in-the-loop oversight that normally mitigates prompt injection risks.
- Ingestion points: The agent processes project tasks that involve reading and editing local files, which may contain untrusted content from external sources.
- Boundary markers: There are no instructions or delimiters provided to help the agent distinguish between authorized user instructions and potentially malicious payloads in the data it is working on.
- Capability inventory: The skill explicitly mentions capabilities such as "write code", "edit notes", and "split tasks", which involve file system modifications.
- Sanitization: No input validation or sanitization requirements are specified for the data the agent is instructed to process autonomously.
Audit Metadata