The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes the official vercel CLI tool through a Python wrapper (scripts/check-deployment.py).
Evidence: The script uses subprocess.run to call vercel inspect and vercel list to retrieve machine-readable deployment metadata.
Security Context: This is the intended primary purpose of the skill. The script includes validation logic to ensure the deployment belongs to the expected project (remotion) and scope, and it explicitly rejects monitoring movable aliases to prevent state confusion.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from the Vercel CLI and user-supplied URLs which could theoretically contain malicious content.
Ingestion points: Deployment metadata retrieved from vercel inspect and URLs provided by users/GitHub PR comments.
Boundary markers: The skill instructions include explicit prompt structures for the heartbeat monitor to delimit context from instructions.
Capability inventory: The skill can execute the vercel CLI and write output to the agent context. It does not have arbitrary file-write or network-exfiltration capabilities beyond interacting with Vercel.
Sanitization: The check-deployment.py script normalizes URLs and deployment IDs, ensuring they conform to Vercel's expected formats before passing them to the CLI.

vercel