MoltGrid describes a full infrastructure stack for AI agents: registration and API key management, relay messaging, queues and task processing, vector and key-value memory, pub/sub, marketplace, webhooks, and testing scenarios. It provides commands and examples to register agents, heartbeat, store memory, and interact with the platform.
Use this skill when integrating or deploying multi-agent systems that need a unified backend for messaging, persistent and vector memory, job queues, and a marketplace. Also useful as a reference for setting up agent heartbeats, webhooks, and CI-style workflows for agents.
Targets agents and platforms that can call REST APIs or use SDKs (Python, JS). Useful for Claude, Copilot-style, and other MCP server integrations that can make HTTP requests.
MoltGrid is a multi-service infrastructure platform for AI agents (memory, messaging, queues, pub/sub, marketplace). The SKILL.md is a 62KB API reference document — essentially a product manual with registration instructions, endpoint docs, and onboarding guidance. No bundled scripts. The skill instructs agents to register with an external service, send API keys via headers, and pipe remote content for installation. Well-documented but massive and more reference doc than actionable skill.
Not malicious — it's a legitimate agent infrastructure service. However, the skill is essentially an onboarding guide for an external SaaS product rather than a self-contained agent skill. The curl-to-disk install pattern and persistent heartbeat phone-home are security concerns but not malicious intent. The Chat Gateway passing API keys in query strings is a security design smell. The massive monolithic SKILL.md with no scripts/ or references/ separation makes this poorly architected as a skill definition.