The Agent Skills Directory

Automated Workflow Approvals: The skill is designed to run examples in an 'auto' mode. To achieve this, it sets several environment variables such as AUTO_APPROVE_MCP=1, AUTO_APPROVE_HITL=1, and SHELL_AUTO_APPROVE=1. These flags are used to bypass interactive confirmation prompts for Model Context Protocol (MCP) tools, Human-in-the-Loop (HITL) requests, and shell command executions during the automated run.
Dynamic Command Execution: The run.sh script executes various pnpm and node commands to build and run example projects. This includes a Node.js heredoc in the rerun_list function that dynamically loads logic from a local file (scripts/run-example-starts.mjs) to manage the list of examples to be executed.
Indirect Prompt Injection Surface: The skill instructions require the AI agent to read example source code and execution logs to validate behavior. While this is a core part of its diagnostic functionality, it creates a surface where instructions embedded within logs or source code could theoretically influence the agent's validation logic.
Process and File Management: The skill manages background processes using PID files and stores execution logs in a .tmp directory within the project root. This is standard behavior for development and testing utilities.

examples-auto-run