autumn-investigate

Pass

Audited by Gen Agent Trust Hub on Sep 3, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill retrieves and processes external log data from API requests and Stripe webhooks, which creates a surface for potential indirect prompt injection attacks.
  • Ingestion points: Log records containing potentially untrusted payloads are accessed via the searchRequestLogs and queryRequestLogs tools.
  • Boundary markers: No explicit delimiters are specified to wrap log data within the agent context.
  • Capability inventory: The skill tools are strictly limited to read-only customer state lookups and log queries (getCustomer, searchRequestLogs, queryRequestLogs).
  • Sanitization: There is no documentation of automated sanitization for the retrieved log content.
  • [SAFE]: No malicious patterns such as credential exfiltration, remote code execution, or obfuscation were detected. The skill follows best practices by using a domain-specific query language and providing structured workflows for investigating specific billing scenarios.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 3, 2026, 05:03 AM
Security Audit — agent-trust-hub — autumn-investigate