polpo-runtime
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The documentation establishes comprehensive security invariants, such as ensuring malformed tool calls are rejected at the runtime boundary and enforcing context trust for retrieved content.
- [COMMAND_EXECUTION]: The reference material for Project Loops explicitly warns developers not to interpolate context into Bash strings, which is a key defense against command injection vulnerabilities.
- [INDIRECT_PROMPT_INJECTION]: The skill documentation details strategies for handling untrusted data, including secret redaction, private-network blocking, and the use of bounded delimiters for external content ingestion in references/context-trust-and-guardrails.md.
- [CREDENTIALS_UNSAFE]: The skill enforces safe credential management by instructing users not to encode secrets in schedule payloads or persistent metadata.
Audit Metadata