The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill creates a webhook listener that accepts HTTP POST requests from the internet via a public tunnel (ngrok or localtunnel). This introduces a surface where external data can be ingested into the agent's context.
Ingestion points: The generated server script listens for data at the POST /webhook endpoint.
Boundary markers: The skill requires HMAC-SHA256 signature verification using an x-nansen-signature header to ensure only authentic Nansen payloads are processed. It also enforces a 1 MB body size limit.
Capability inventory: The skill uses Write to create the server script and Bash to execute the server, tunnels, and JSON parsing commands.
Sanitization: While the script validates the signature and parses JSON, the contents of the alerts are logged directly to stdout without further sanitization, which is a common vector for indirect injection attacks.
[COMMAND_EXECUTION]: The skill uses node -e (evaluate) for utility operations such as generating a random cryptographic secret and parsing JSON output from a local API.
Evidence: node -e "console.log(require('crypto').randomBytes(32).toString('hex'))" and curl -s http://127.0.0.1:4040/api/tunnels | node -e "...".
[DYNAMIC_EXECUTION]: The skill generates a complete Node.js server script (nansen-webhook-server.mjs) from a template provided in the instructions. This script incorporates user-generated secrets and environment variables at runtime.
[EXTERNAL_DOWNLOADS]: The skill utilizes npx localtunnel, which downloads and executes the localtunnel package from the npm registry. It also provides instructions for installing ngrok via Homebrew.

nansen-alerts-webhook-listener