The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill implements a local Express.js receiver that ingests enrichment data from external Clay webhooks, creating a potential surface for indirect prompt injection if the agent processes the logged output.
Ingestion points: The /api/clay/enriched endpoint in src/clay-receiver.ts accepts JSON payloads containing data from external enrichment sources.
Boundary markers: Absent. The skill does not provide specific delimiters or instructions for the agent to ignore instructions embedded within the webhook data.
Capability inventory: The skill has access to Bash, Write, and Edit tools. The provided scripts perform console logging and standard HTTP POST requests.
Sanitization: Absent. The receiver logs fields from the incoming request body directly to the console without sanitization.
[COMMAND_EXECUTION]: The instructions guide the user to execute standard development commands such as npm, npx, and ngrok to manage dependencies, run TypeScript scripts, and establish local tunnels.
[DATA_EXFILTRATION]: The skill utilizes ngrok, a well-known service, to expose a local port (localhost:3000) to the public internet. This enables cloud-based Clay columns to send data back to the developer workstation. The skill includes explicit warnings to avoid tunneling production data, live secrets, or CRM credentials through this setup.
[EXTERNAL_DOWNLOADS]: Recommends the installation of standard development packages and utilities, including express, ngrok, tsx, and vitest, from the official npm registry.

clay-local-dev-loop