share-of-voice-tracker
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest data from external sources such as web scrapes, user exports, and news feeds, which presents a surface for indirect prompt injection.\n
- Ingestion points: Data ingestion points include user-provided exports and scrapes from public listening connectors as described in
SKILL.mdandreferences/connectors.md.\n - Boundary markers: The
references/skill-contract.mdfile explicitly mandates that external content be treated as evidence, not instructions, and directs the agent to ignore any commands embedded within the data.\n - Capability inventory: The skill is capable of interacting with various marketing platforms (e.g., Google Ads, HubSpot) but strictly requires explicit user authorization before performing any persistent writes or external side effects.\n
- Sanitization: The framework enforces evidence labeling (e.g., measured, estimated) and relies on user-authorized manual approval gates for sensitive actions.\n- [SAFE]: The skill instructions emphasize security best practices, including the prohibition of hardcoded API keys and the use of read-only defaults for data connectors.
Audit Metadata