Show HN: Pantheon MCP – a central server for AI agent definitions (pantheon-mcp.com)

Pantheon MCP is a lightweight central server for storing and serving AI “agent” definitions — not just a flat index but a curated “pantheon” of reusable agent personalities, capabilities and instruction sets. Announced as an npm-distributable tool (npm install -g pantheon-mcp@latest) the server is intended to be referenced directly from agent prompts (“[insert task here] using agentic instructions from pantheon-mcp”) and integrated into agent platforms via a simple stdio-backed MCP configuration (example JSON shows an mcpServers entry that runs `npx pantheon-mcp`). For the AI/ML community this matters because a shared, discoverable registry of agent definitions enables reuse, composability and reproducibility of agentic behaviors across projects. Technically it’s designed for easy integration (stdio transport, npx invocation), making it simple to plug into existing multi-agent controllers or orchestration layers and standardize how agent roles and prompts are distributed. Implications include faster prototyping of agent ecosystems, clearer governance/versioning of agent specs, and easier benchmarking or A/B testing of agent personalities in production pipelines.