apollo-lead-finder
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill retrieves and processes data from the Apollo.io API, which is an external source and could potentially be manipulated by a third party to include prompt injection payloads in lead names or titles. \n
- Ingestion points: Data enters the context in
scripts/apollo_lead_finder.pythrough search and enrichment API responses. \n - Boundary markers: The skill structures the fetched data into JSON manifests and CSV files, separating it from the instructional flow. \n
- Capability inventory: The skill possesses database write permissions (to the
peopletable in Supabase) and local file write permissions (to theoutput/directory). \n - Sanitization: The instructions enforce a human-in-the-loop policy, requiring the agent to present samples to the user and obtain explicit approval before performing enrichment (which costs credits) or upserting data to the database. This significantly mitigates the risk of automated obedience to injected instructions.
Audit Metadata