plaid-fintech
Pass
Audited by Gen Agent Trust Hub on Nov 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODEMETADATA_POISONING
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill directs the agent to ingest and process data from the Plaid API, including transaction syncs and webhooks. • Ingestion points: Plaid API responses and webhook notifications (SKILL.md). • Boundary markers: None defined in the instructions to isolate external data from the agent prompt. • Capability inventory: No scripts or tool permissions are defined in the skill, but the instructions assume the agent has the capability to perform API calls and sync transactions. • Sanitization: No specific data validation, filtering, or sanitization logic is provided.
- [NO_CODE]: The provided skill file is purely documentation and does not contain any functional scripts or automated tool configurations.
- [METADATA_POISONING]: The 'Sharp Edges' section contains a table with placeholder text and high severity labels without describing actual issues, which may be misleading.
Audit Metadata