feedsmith
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill provides instructions for managing and verifying the
feedsmithpackage through standard package managers. - Evidence includes usage of
bun add feedsmith,bun info feedsmith, andnpm view feedsmithto ensure correct versioning and installation. - [INDIRECT_PROMPT_INJECTION]: The skill acts as a guide for parsing external web feeds, creating an ingestion point for untrusted data.
- Ingestion points: The documentation (specifically
references/parsing.md) describes how to parse external content viaparseFeedand dedicated format parsers. - Boundary markers: There are no specific instructions for setting up prompt boundaries to isolate the data from the agent's instructions.
- Capability inventory: The skill instructions are limited to data parsing and code structure planning; they do not grant the agent capabilities for sensitive file system modifications or unauthorized network activity based on the feed content.
- Sanitization: The skill mentions specific parsing behaviors for Atom xhtml (stripping wrappers, escaping entities) which serve as a rudimentary form of data sanitization.
- [NO_CODE]: The skill consists entirely of markdown documentation and configuration files, containing no executable scripts or binary files within its distribution.
Audit Metadata