cash-flow-snapshot

Pass

Audited by Gen Agent Trust Hub on Sep 20, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests data from multiple external sources including QuickBooks, PayPal, Stripe, Square, and CSV files, which could be leveraged to provide malicious instructions to the agent.
  • Ingestion points: Defined in SKILL.md (Step 2) as various financial connectors and local CSV files.
  • Boundary markers: The skill instructions recommend verifying CSV column mappings and using authorized data but do not specify delimiters or warnings to ignore instructions within the data content.
  • Capability inventory: The skill utilizes sandbase_call_tool (network read) and a host spreadsheet capability (file write) to perform its functions.
  • Sanitization: There are no descriptions of sanitization or filtering for text fields retrieved from external APIs or files.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 20, 2026, 08:11 AM
Security Audit — agent-trust-hub — cash-flow-snapshot