The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's workflow requires the agent to ingest and analyze arbitrary content from the local repository, such as source code, tests, and documentation files (e.g., AGENTS.md, PRODUCT.md, CONTEXT.md). This process exposes the agent to potential indirect prompt injection if those files contain adversarial instructions designed to subvert the agent's goals. * Ingestion points: Local repository files, documentation, and source code accessed during the 'Inspect' and 'Clarify' phases. * Boundary markers: The instructions do not specify the use of delimiters or warnings to ignore instructions embedded within the processed data. * Capability inventory: The skill assumes the agent has access to terminal tools for git operations, testing frameworks (e.g., npm test), and build systems. * Sanitization: There are no instructions for sanitizing or validating the content of the files before processing.
[EXTERNAL_DOWNLOADS]: The documentation references external GitHub repositories provided by community contributors as optional resources for deeper process guidance. These references point to well-known community resources on GitHub.
[COMMAND_EXECUTION]: The workflow involves executing standard development tools such as git, package managers, compilers, and linters. The skill emphasizes that 'Nothing leaves the machine without a go,' requiring explicit human authorization for external actions like pushing code or merging branches.

workflow