The Agent Skills Directory

[EXTERNAL_DOWNLOADS] (LOW): The documentation and scripts recommend installing the netlify-cli package from npm. This is a standard industry tool, but it represents an external dependency outside of the predefined trusted organization list.
[COMMAND_EXECUTION] (LOW): The scripts/test_function_locally.sh script utilizes the eval command to execute a dynamically built curl string. This pattern is vulnerable to command injection if input parameters such as the method or data payload are sourced from untrusted entities.
[DATA_EXFILTRATION] (LOW): The scripts/check_deployment.sh script includes a verbose mode that outputs all site environment variables to the console. This functionality could lead to the exposure of production secrets if the script's execution logs are captured by unauthorized processes or users.
[PROMPT_INJECTION] (LOW): Several webhook handler templates (e.g., assets/examples/api-route.ts) ingest untrusted external data, forming an indirect prompt injection surface. Evidence: (1) Ingestion Points: Request bodies are read directly from external HTTP POST requests in api-route.ts. (2) Boundary Markers: HMAC signature validation is demonstrated but explicitly marked as optional or requiring implementation in templates. (3) Capability Inventory: Handlers have the capability to trigger further network requests via fetch. (4) Sanitization: Input is parsed as JSON without any filtering for malicious instructional content that could affect downstream AI processes.

netlify-integration