sub-agents-mcp and director
One tool defines task-specific AI sub-agents in Markdown for any MCP-compatible tool, while the other provides MCP playbooks for AI agents, making them complementary tools within the MCP ecosystem where the former defines the agents that the latter orchestrates via playbooks.
About sub-agents-mcp
shinpr/sub-agents-mcp
Define task-specific AI sub-agents in Markdown for any MCP-compatible tool.
Markdown-defined agents execute via pluggable CLI backends (Cursor, Claude Code, Gemini, or Codex), with built-in session management and error handling for multi-step workflows. The MCP server architecture enables teams to define reusable agents once and consume them across any MCP-compatible IDE or tool, creating portable Claude Code–style sub-agent capabilities outside Claude's environment.
About director
director-run/director
MCP Playbooks for AI agents
Provides a gateway that bridges AI agents with MCP servers through composable "playbooks"—collections of tools, prompts, and configuration that can be filtered, isolated, and swapped dynamically. Implements centralized OAuth management, tool filtering for context preservation, and unified JSON logging across all connected clients (Claude, Cursor, VSCode). Runs as a local service exposing playbooks through a single MCP endpoint, compatible with any MCP-compliant client or server.
Related comparisons
Scores updated daily from GitHub, PyPI, and npm data. How scores work