AgentID

AgentID is a coding tool. AgentID connects Claude Code, Cursor, Codex and other MCP clients to one hub that gives each agent a persistent profile, shared memory, connectors, files, email and live monitoring, with a library of 180+ prebuilt agents.

AgentID is a management layer for AI agents. Instead of configuring memory, tools and credentials separately inside every coding assistant or chatbot, a user creates an agent once and connects it to whichever client it runs in. The agent receives a persistent profile (name, voice, beliefs, skills and rules), shared memory, a connector catalog, file storage, its own email address and a key vault, and everything it does is reported to a live dashboard. The product is aimed at builders who run several agents across different environments, such as Claude Code, Cursor, Windsurf, Codex, Copilot, Gemini or any client that speaks MCP. Connection works through an MCP server URL: the user creates an agent, copies its URL and pastes it into the client with an authorization token. A system prompt export covers setups where MCP is not available. Skills, which are reusable capability packs written in markdown, can be attached to an agent and are injected into the system prompt it exports. In practice the dashboard is organised around four views. The floor shows agents working or idle in real time, the roster lists runs, tokens, status and platform, the memory view shows facts that agents write and share across tools and sessions, and the tasks view handles missions and handoffs between agents. A Drive holds files, API keys, skill packs and connectors, with each agent limited to what it has been approved to access. A library of more than 180 prebuilt agents, grouped by discipline such as engineering, design, marketing, sales and finance, gives new users a starting point they can customise. Among alternatives, AgentID sits closer to an agent operations and identity hub than to an agent builder or an LLM framework. It does not replace the coding assistants it connects to; it gives them continuity and oversight. The homepage describes a free tier with a short Pro trial, which makes it easy to test, though the depth of the monitoring depends on how faithfully each connected client reports its activity.