share-of-voice-tracker
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to fetch and analyze content from several external platforms, creating a surface for instructions embedded in social posts or news articles to influence the agent's behavior.
- Ingestion points: Data is ingested through multiple connectors including
hn.py,bluesky.py,fediverse.py,discourse.py,gdelt.py,tavily.py, and user-provided exports. - Boundary markers: The instructions include a specific warning to treat every fetched article and post as untrusted input per the skill's security guidelines, though explicit prompt delimiters for the mention text are not defined in the provided file.
- Capability inventory: The skill has the capability to write trend data and panel records to internal storage (
memory/social/share-of-voice-tracker/) and can propose updates to a channel registry viaregistry-events.py. - Sanitization: While the skill mandates an 'untrusted' posture, it does not detail specific programmatic sanitization or filtering logic applied to the external mention text before it is processed for sentiment or volume calculation.
Audit Metadata