podium-contact-dedup
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests contact data from the external Podium API, creating a potential surface where an attacker-controlled record could contain instructions designed to manipulate the agent during the data review process.\n
- Ingestion points: Contacts are fetched from the Podium API (
api.podium.com) viascripts/find_duplicates.pyand stored in a local SQLite database.\n - Boundary markers: The skill does not provide explicit delimiters or instructions to the agent to ignore potentially malicious text within contact fields when reviewing cluster reports or audit logs.\n
- Capability inventory: The skill utilizes
Writeaccess to the local filesystem,Bashexecution for Python and SQLite operations, and network access to interact with the Podium API.\n - Sanitization: The implementation uses standard data safety measures such as parameterized SQL queries and JSON serialization, which protect the processing scripts but do not fully isolate the LLM from text-based prompt injection within the data.
Audit Metadata