A curated directory of agent products and open-source templates. Find the agent that fits your workflow, then extend it with MCP servers from this site.
Anthropic
Anthropic’s agentic workspace for knowledge work.
WorkAny
Local desktop AI agent for everyday office work.
Tencent
Tencent’s multi-agent workspace for everyday office work.
Common questions about agent products, templates, and how they relate to MCP.
An AI agent is a system built on a large language model that can plan multi-step work and execute it with tools — writing code, browsing the web, managing files, or talking to external services — rather than only answering in text.
An agent product is a ready-to-use application or service, such as a coding agent or a general-purpose assistant. An agent template is a starter codebase or workspace — for example an OpenClaw starter — that you run and customize yourself to build your own agent.
MCP (Model Context Protocol) is the open standard agents use to connect to external tools and data sources. Most agents listed here support MCP, which means every MCP server in this directory can extend what they are able to do.
Use the feedback button on any page to suggest an agent product or template, including a link to its website or repository. Curated suggestions are reviewed and added regularly.
Start with the work you want to delegate: coding, research, writing, customer support, or automation. Then compare the agent's supported tools, MCP compatibility, deployment model, privacy controls, pricing, and permission requirements. Try it first with a small, reversible task.
Use trusted MCP servers, grant the smallest set of permissions needed, and keep API keys in a secure secret store. Start with read-only or sandbox access, review the agent's planned tool calls, and require approval before it performs actions that change data or reach external services.
