clay-reference-architecture
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The architecture is designed to ingest and process data from untrusted external sources such as webhooks, CRM imports, and CSV uploads, which is then used for AI-driven personalization.
- Ingestion points: Identified in
SKILL.md(Step 1 Pattern A/B/C) andreferences/implementation-guide.md(Architecture Diagram). - Boundary markers: No explicit delimiters or instructions to ignore embedded instructions are present in the provided architecture guidelines.
- Capability inventory: The integration code uses network capabilities to interact with external APIs (
api.clay.com,api.hubapi.com). - Sanitization: A
validators.tsfile is suggested in the project structure withinreferences/implementation.md, but no specific validation or sanitization logic is implemented in the code examples. - [DYNAMIC_EXECUTION]: The troubleshooting section includes commands for executing inline TypeScript code to test internal services.
- Evidence:
npx ts-node -eis used inreferences/implementation.mdto run snippets that resolve components from a dependency injection container.
Audit Metadata