The Agent Skills Directory

[PROMPT_INJECTION]: The skill identifies a standard indirect prompt injection attack surface by processing user-provided arguments to modify monorepo configurations.
Ingestion points: User requests are captured via the $ARGUMENTS variable in command/turborepo.md for processing.
Boundary markers: The skill utilizes <user-request> tags to encapsulate untrusted data, providing a clear delimiter between instructions and user-provided input.
Capability inventory: The agent is guided to modify critical configuration files such as package.json and turbo.json across the workspace.
Sanitization: There are no explicit instructions for the agent to sanitize or validate the user-provided request before applying changes to the configuration files.
[COMMAND_EXECUTION]: The skill provides instructions for executing standard development and build orchestration commands.
Runs turbo run <tasks> to manage parallel task execution and individual package caching.
Utilizes npx turbo-ignore to optimize CI workflows by detecting unchanged packages and skipping unnecessary work.
Employs turbo watch for automated development loops and dependency-aware task restarts.
[EXTERNAL_DOWNLOADS]: The skill references and recommends tools and services from well-known providers and official sources.
Directs users to official documentation and schemas at turborepo.dev and vercel.com.
Recommends common community utilities such as syncpack, manypkg, and sherif for managing workspace dependencies.
Provides guidance on configuring Vercel Remote Caching to accelerate builds across environments.

turborepo