authoring-scouts
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill serves as a technical manual and framework for authoring autonomous agents. It belongs to the official vendor namespace and describes intended platform functionality.
- [INDIRECT_PROMPT_INJECTION]: The skill describes a platform where agents ('scouts') process data from external sources like GitHub issues, social feeds, and warehouse tables. The documentation includes a dedicated security section ('Safety: treat ingested content as untrusted data') that instructs authors to sanitize content, avoid following instructions embedded in data, and use paraphrasing to prevent prompt injection and PII leakage.
- [COMMAND_EXECUTION]: The scouts described can execute sandboxed CLI tools (e.g., git, rg) for data analysis. This is a core part of the feature's design and is restricted by platform-level security controls such as domain allowlists and sandboxing.
- [EXTERNAL_DOWNLOADS]: The skill documentation notes that scouts can fetch data from external APIs or repositories. This behavior is expected for the use case and is managed through a configurable network access policy that defaults to trusted domains.
Audit Metadata