bank-sync
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external financial data from the Plaid API, creating a vulnerability surface where malicious instructions could be embedded in transaction descriptions to influence the behavior of the AI agent tools.\n
- Ingestion points: Transaction data is imported from the Plaid API during the initial and daily sync workflows defined in
SKILL.md.\n - Boundary markers: The instructions do not define boundary markers or include explicit directives to ignore instructions contained within the imported transaction data.\n
- Capability inventory: The skill uses the
categorizeandtransaction_searchtools to analyze and process the imported transaction records.\n - Sanitization: There is no evidence of sanitization, filtering, or validation of transaction descriptions before they are processed by the agent.
Audit Metadata