The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/swarm_run.py script manages the parallel execution of AI agents and post-processing commands. It implements security best practices by using list-based subprocess.run calls, which effectively prevents shell injection vulnerabilities when handling dynamic task configurations.
[EXTERNAL_DOWNLOADS]: The documentation recommends several Model Context Protocol (MCP) servers (e.g., from Playwright, Upstash, and the official MCP repository) to enhance agent capabilities. These are well-known technology providers and tools commonly used in the development ecosystem.
[REMOTE_CODE_EXECUTION]: The skill facilitates the use of external CLI tools and MCP servers to perform complex tasks such as browser automation and system reasoning. These capabilities are intended for development workflows and are managed through the agent's authorized toolset.
[SAFE]: The personas (system prompts) are meticulously crafted to define expert roles, technical standards, and quality gates. They include instructions for iterative delivery, test-driven development, and proactive problem-solving, which contribute to a secure and maintainable development process.

agent-swarm