Getting started

FindAgent documentation

FindAgent is the cross-LLM AI agent marketplace. Discover focused AI agents and put them to work inside the LLM client you already use — Claude, ChatGPT, Gemini, Cursor, and more.

What FindAgent is

Every agent on FindAgent is a packaged, reusable capability: a system prompt, an optional set of tools, and example flows — published by a creator and ready to drop into your LLM client.

Some agents are recipes (a focused prompt and example flows). Others are “doer” agents that actually take action — pulling a GA-style analytics report, generating a monthly tax spreadsheet, or managing a storefront — by calling real tools on your behalf.

The catalogue holds a few shapes of agent:

Recipe agents — a packaged system prompt and example flows.
Tool-using (doer) agents — agents that call real tools (APIs, data sources) through a shared, trusted runtime.
Skills bundles — a Claude, Cursor, Windsurf, or Continue skills-or-rules repo packaged as one agent, with each skill served as an MCP prompt.
Code agents — a creator's real code from a GitHub repo, run 1:1 in an isolated sandbox after review.
MCP servers — a remote MCP server someone already hosts, listed in the MCP directory so any MCP client can discover and connect it.

You browse the catalogue on two axes — Industry (the vertical an agent serves) and Discipline (the kind of work it does, like Customer Service or Software Development). Lead with whichever matches how you think, then intersect the two from a category hub.

What a "doer" agent is

A doer agent declares tools in its manifest. Each tool describes what it does and how it is invoked (an HTTP call, a prompt template, or a call into another agent). A shared, trusted runtime executes those declared tools — the agent never ships executable code. This keeps the catalogue safe to install while still letting agents do real work. See the security model for why this matters.

Two ways to use an agent

There are two delivery paths, and most agents support whichever your client prefers:

Connect (hosted) — the primary path. The agent runs on FindAgent and is reachable over MCP. You add its server URL — like https://mcp.findagent.cloud/agents/<slug> — as a remote connector in your client and sign in once. Nothing to download, nothing to keep running locally, and it works from web and mobile clients too. Connect guide →
Install (local). Prefer to run on your own machine? Install the agent into your LLM client with the FindAgent CLI or the guided web installer; tools execute through a local runtime, and any credentials stay on your device. Install guide →

Which should I use?

Connecting to the hosted version is the recommended path — an agent reachable from any client (including web and mobile) with nothing to maintain. Install locally instead if you'd rather keep everything — including credentials — on your own machine.

Go further

Combine several agents into a team with an agent-to-agent topology and run them as one unit — see Departments.
Publishing your own agent? Start with the creator guide and the manifest spec. You can even submit straight from your AI client — see MCP & API.
Looking for a remote MCP server? Browse the MCP directory and connect any listing to your client.
Integrating programmatically? See the API & integration reference.

Browse the marketplace