The Agent Skills Directory

[SAFE]: The skill provides standard documentation and architectural templates for building integrations with the Instantly.ai service. It follows best practices by recommending the use of environment variables for API keys and providing clear project structure.
[INDIRECT_PROMPT_INJECTION]: The architecture defines components for ingesting external data (e.g., CSV imports, CRM synchronization). This introduces an attack surface where malicious input could influence agent behavior if not properly sanitized.
Ingestion points: Data enters via src/leads/import.ts (CSV/CRM import).
Boundary markers: Not present in the provided architectural snippets.
Capability inventory: The skill design involves network requests (fetch), file operations (Write, Edit), and package management (Bash).
Sanitization: While the code uses encodeURIComponent for URL parameters, broader sanitization for imported data is left for implementation, though the instructions recommend human-in-the-loop and validation gates.

instantly-reference-architecture