The Agent Skills Directory

[COMMAND_EXECUTION]: The skill facilitates the execution of Prisma Platform CLI tools using package runners like npx and bunx. Specifically, it uses @prisma/cli and create-prisma to handle application deployment, project scaffolding, and environment management.
[EXTERNAL_DOWNLOADS]: The skill's workflow involves downloading the latest versions of official Prisma packages from the NPM registry at runtime. These include @prisma/cli, create-prisma, and @prisma/compute-sdk.
[DATA_EXFILTRATION]: Includes a utility script scripts/smoke-deployed-app.mjs designed to perform HTTP requests to verify the availability of deployed applications. The script implements security controls to prevent Server-Side Request Forgery (SSRF) by blocking requests to loopback and localhost addresses by default.
[CREDENTIALS_UNSAFE]: The instructions demonstrate high security awareness by explicitly forbidding the agent from printing or exposing sensitive values such as PRISMA_SERVICE_TOKEN, DATABASE_URL, or other environment secrets found in .env files or CLI output.
[PROMPT_INJECTION]: While the skill uses non-interactive flags (--no-interactive) and structured output (--json) to facilitate automated processing by the agent, this is directed at improving technical interoperability rather than concealing malicious actions from the user.

prisma-compute