The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides code templates for a webhook receiver that ingests untrusted external data.\n
Ingestion points: The api/clay/callback.ts template processes data from req.body.\n
Boundary markers: The implementation includes signature validation using crypto.timingSafeEqual and x-clay-signature headers to verify the sender.\n
Capability inventory: The skill possesses deployment capabilities via Bash(vercel:*), Bash(gcloud:*), and Bash(docker:*) tools, along with file system access via Write and Edit.\n
Sanitization: While the template validates the source, it does not demonstrate explicit sanitization of the payload content before processing, which is a standard surface for indirect prompt injection if the data is later used in LLM prompts.\n- [COMMAND_EXECUTION]: Utilizes Bash tools for legitimate deployment operations.\n
Executes vercel --prod for serverless production deployments.\n
Executes gcloud builds submit and gcloud run deploy for managing containerized applications on Google Cloud Platform.\n
Executes docker compose for container orchestration workflows.\n- [EXTERNAL_DOWNLOADS]: Interacts with well-known external services for status checks and deployment activities.\n
Fetches health status data from Clay's official domain (app.clay.com).\n
Downloads and interacts with deployment assets through Vercel and Google Cloud official services.

clay-deploy-integration