stripe-integration-expert
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The
integration_auditor.pyandwebhook_validator.pyscripts ingest and process data from user-provided project directories, which constitutes an indirect prompt injection surface. - Ingestion points: The scripts read the content of all source files within a specified
project_dirprovided as a command-line argument to the auditor and validator tools. - Boundary markers: Absent. The scripts read raw file content into memory for pattern matching without specific delimiters or instructions that would prevent an LLM from interpreting instructions embedded within the scanned code.
- Capability inventory: The skill includes Python scripts that perform directory traversal and file reading. It also provides reference code for network-based webhook handlers and checkout sessions. The scripts do not execute the code they scan or perform network operations based on the scanned content.
- Sanitization: The scripts use standard file reading practices with
errors="replace"to safely handle non-UTF-8 characters during the scanning process.
Audit Metadata