The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill communicates with Vercel's official APIs using the Vercel CLI to retrieve deployment metrics, usage statistics, and project configurations. These communications are directed to a well-known service and are necessary for the skill's core functionality.
[INDIRECT_PROMPT_INJECTION]: The skill analyzes the target application's codebase and external metrics, creating an ingestion surface for potentially untrusted data. 1. Ingestion points: scripts/scan-codebase.mjs (reads the local project files) and scripts/collect-signals.mjs (collects API signals). 2. Boundary markers: The skill uses explicit investigation briefs produced by lib/investigation-brief.mjs which contain clear instructions to the AI agent to treat all ingested data as evidence and never as instructions. 3. Capability inventory: The skill can read/write files in its run directory and execute specific CLI tools using execFile. 4. Sanitization: The lib/vercel.mjs module implements redactSensitiveText to ensure that authentication tokens and private IDs are removed from output before processing.
[COMMAND_EXECUTION]: The skill performs automated execution of the vercel CLI and its own internal logic scripts. It uses execFile rather than exec, which mitigates the risk of shell injection by passing arguments as an array rather than a single string.

vercel-optimize