sus-research
Pass
Audited by Gen Agent Trust Hub on Aug 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from untrusted external sources, creating a surface for indirect prompt injection attacks.
- Ingestion points: The skill reads content from various external sources including official documentation, APIs, websites, research studies, user reviews, and surveys.
- Boundary markers: The instructions include strong guidelines for recording provenance (URLs, dates, versions) and evidence tracking, which helps contextualize external data, though it does not explicitly instruct the agent to ignore hidden commands within those sources.
- Capability inventory: The skill has the capability to write research artifacts to a specific local directory (
~/.agents/artifacts/) and perform file deletion/cleanup during the closing phase. - Sanitization: While the skill mandates "cutting structural bloat" and "rerunning checks," it does not describe specific sanitization or escaping of external content before processing.
- [DATA_EXPOSURE]: The skill manages files in a specific local directory (
~/.agents/artifacts/). It includes safety checks to prevent overwriting repository files or vendor directories, which is a positive security practice for artifact management.
Audit Metadata