The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes standard CLI tools including curl for API interaction, npm for package installation, and npx for Prisma migrations and script execution (tsx). These operations are required for the stated purpose of project setup.
[CREDENTIALS_UNSAFE]: The skill handles sensitive service tokens and database connection strings. It correctly instructs the agent to use environment variables and .env files for storage, and specifically warns the user to ensure .env is listed in .gitignore to prevent accidental exposure.
[EXTERNAL_DOWNLOADS]: The skill installs necessary dependencies (prisma, @prisma/client, @prisma/adapter-pg, pg, dotenv) from the official npm registry. These are well-known, legitimate packages from the Prisma ecosystem.
[DATA_EXFILTRATION]: Network requests are directed to the official Prisma API domain (api.prisma.io) for resource provisioning. This is the intended and necessary behavior for a management tool associated with the vendor 'prisma'.
[INDIRECT_PROMPT_INJECTION]: The skill features a surface for indirect prompt injection in Step 6, where it generates a database schema based on natural language descriptions provided by the user. However, this is a core intended feature, and the generated content is reviewed by the user before being applied via migration.
Ingestion points: User-provided schema descriptions in Step 6.
Boundary markers: None explicitly defined in the prompt interpolation.
Capability inventory: File writes to schema.prisma, npx prisma migrate execution, and curl API calls.
Sanitization: None specified for the natural language processing step, relying on agent-level safety filters.

prisma-postgres-setup