# MCP server > Connect Claude Code, Codex, Cursor, VS Code or any MCP client to Fous, with tools to find, run and build methods on any website, manage workflows and connect accounts. Source: https://docs.fous.com/guides/mcp Index: https://docs.fous.com/llms.txt Connect any MCP client to `https://api.fous.com/mcp` (Streamable HTTP), with your [API key](https://docs.fous.com/authentication.md) as a bearer header. Or set the key as `FOUS_API_KEY` and tell your agent `set up https://fous.com/SKILL.md`: it connects the server and saves the skill. The MCP server and API keys work on every plan ([plans](https://docs.fous.com/guides/billing.md#plans)). On Free, an MCP client runs and reads your methods; the `build` tool answers `403 plan_limit`, and builds start in [Studio](https://app.fous.com). ## Connect your client ### Claude Code Adds Fous for every project, reading the key from your shell environment. ```bash claude mcp add --scope user --transport http fous https://api.fous.com/mcp \ --header "Authorization: Bearer ${FOUS_API_KEY:?Set FOUS_API_KEY to your Fous API key}" ``` Reference: https://code.claude.com/docs/en/mcp ### Codex Codex shares this configuration between its CLI and IDE extension. ```bash codex mcp add fous --url https://api.fous.com/mcp --bearer-token-env-var FOUS_API_KEY ``` Reference: https://developers.openai.com/codex/mcp/ ### Cursor For every project. Use .cursor/mcp.json in a repository to scope it to one project. File: `~/.cursor/mcp.json`. ```json { "mcpServers": { "fous": { "url": "https://api.fous.com/mcp", "headers": { "Authorization": "Bearer ${env:FOUS_API_KEY}" } } } } ``` Reference: https://cursor.com/docs/mcp ### VS Code VS Code asks for the key once and keeps it out of the file. File: `.vscode/mcp.json`. ```json { "inputs": [ { "type": "promptString", "id": "fous-api-key", "description": "Fous API key", "password": true } ], "servers": { "fous": { "type": "http", "url": "https://api.fous.com/mcp", "headers": { "Authorization": "Bearer ${input:fous-api-key}" } } } } ``` Reference: https://code.visualstudio.com/docs/copilot/customization/mcp-servers ### Any MCP client Streamable HTTP with a Bearer header. The add-mcp helper writes most clients’ config. ```bash npx -y add-mcp https://api.fous.com/mcp -g -n fous \ --header "Authorization: Bearer ${FOUS_API_KEY:?Set FOUS_API_KEY to your Fous API key}" ``` Reference: https://modelcontextprotocol.io/clients ### A Claude Code project Commit this `.mcp.json`. Each person's own `FOUS_API_KEY` fills the header. ```json { "mcpServers": { "fous": { "type": "http", "url": "https://api.fous.com/mcp", "headers": { "Authorization": "Bearer ${FOUS_API_KEY}" } } } } ``` ### Your own agent, server side An agent that calls tools with no person approving them should use `?read_only=true`: no `run_action`, `build` or `stop`, and the other tools only read (`connections` lists and reads connections; `workflows` lists and reads workflows and their runs, without their webhook addresses). Anthropic Messages API, with `anthropic-beta: mcp-client-2025-11-20`: ```json { "mcp_servers": [ { "type": "url", "url": "https://api.fous.com/mcp?read_only=true", "name": "fous", "authorization_token": "" } ], "tools": [{ "type": "mcp_toolset", "mcp_server_name": "fous" }] } ``` OpenAI Responses API: ```json { "tools": [ { "type": "mcp", "server_label": "fous", "server_url": "https://api.fous.com/mcp?read_only=true", "authorization": "", "require_approval": "never" } ] } ``` ## The tools - `search`: Find your organization’s methods for a plain-language need, with their input schemas. - `run`: Run a method that reads data, with inputs that fit its schema. - `run_action`: Run a method that changes something on a website (posts, books, sends), only when the user asked for it. - `get_run`: The result of a slow run (run_…) or a workflow run (wfr_…), or the status of a build (bld_…). - `stop`: Stop a run (run_…), a build (bld_…) or a workflow run (wfr_…) that is still going. A run or build costs nothing then; a workflow run keeps the cost of the steps that had finished. - `build`: Build a method that reads or does something on a website from a plain-language request, with up to 4 images (a link or base64); with one of your methods, change it; with a build_id, send that build a message. - `connections`: Connect a website account (each step the site asks for comes back for the person’s answers, with a Studio link to sign in there instead), list your organization’s connections, check one, make one its site’s default, or disconnect one. - `methods`: List your organization’s methods, read one (input and output schemas, an example, versions), switch the version its calls run, or delete one. - `builds`: Your organization’s builds of the past 30 days, newest first; dismiss one that ended, or fetch a picture sent with a build or a message. - `workflows`: List your organization’s workflows, read one or its runs, create one from a prompt or a definition, change it, build the method a step waits for or remove a step, turn it on or off, delete it or run it. - `account`: Your organization, plan, credits left, limits, builds and prices, as GET /v1/account. Each tool does what its [endpoint](https://docs.fous.com/api-reference.md) does: `methods` as in [Find and manage methods](https://docs.fous.com/guides/methods.md), `run` and `run_action` as in [Run a method](https://docs.fous.com/guides/runs.md), `build` as in [Build a method](https://docs.fous.com/guides/builds.md). `methods`, `builds`, `workflows` and `connections` take an `action` that picks the operation. `structuredContent` is what the API returns; a run's `data` matches the method's `output_schema`. The text next to it is a one-line summary and JSON shortened to fit 24,000 characters; in a run's text, a list of two or more records is written as `{columns, rows}` (each field name once, `null` for a field an item lacks) to save tokens. ## Slow work After about 50 seconds, `run` and `run_action` return `status: "running"` and an `id`. Call `get_run` with it, or `stop`, instead of sending it again. `build` returns at once (send `wait`, up to 60 seconds, to hold the answer); follow it with `get_run`. `get_run` and `stop` take a run (`run_…`), a build (`bld_…`) or a workflow run (`wfr_…`). `get_run` answers a run that failed with `status: "failed"` and its `error`, and `stop` with the run or build as it now stands (a method run in progress is free; a workflow run's finished steps stay charged). ## Workflows `workflows` runs the [workflows](https://docs.fous.com/guides/workflows.md) API from one tool, with the same checks, through `action`: | `action` | What it does | | --- | --- | | `list` | Your organization's workflows. | | `get` | One workflow by `id` (`wf_…`): its `definition`, what it still needs filled in (`incomplete`) and what one run costs (`cost`). | | `runs` | The runs of `id`, newest first (`limit`, `before`). | | `create` | From `prompt` (the AI drafts it from your methods and saves it turned off; `timezone` sets when "9am" is) or from `name` and `definition`. When the AI asks something back, `workflow` is `null` and `ai` has `reply` and `question`: answer with `messages`, the whole conversation so far. | | `update` | Change `id` with `prompt` or `messages` (the AI changes it, saved like any change; `ai` says what it did), or with `name`, `description` or `definition`. | | `build_step` | Build the method of a step that is waiting for one: `id`, `step` (its key), and `site` when the step names none. The step uses the method once it is built; follow it with `get_run`. | | `remove_step` | Take `step` out of `id`. | | `enable`, `disable` | Turn `id` on or off. | | `delete` | Delete `id`; its runs still going are canceled. | | `run` | Run `id` with `input` for its trigger's inputs. `wait` (0 to 60, 30 by default) holds the answer. | A run is a workflow run with a `wfr_` id. It answers with its steps once it has ended; while it is still going it answers `status: "queued"`, `"running"` or `"waiting"`: call `get_run` with its id to wait for it, or `stop` to cancel it. On the read-only server, `workflows` has only `list`, `get` and `runs`. ## Actions and sign-ins - `run` only reads. A `write` method goes through `run_action`, marked destructive so the host can ask you first. So are `build`, `stop`, `workflows`, `methods` and `connections`, so the host can ask before it changes something. - On `connection_required`, give the person `connect_url`. Or call `connections` with `action: "connect"` and `site`, and relay each step to the person until the connection is `active` ([Connect an account](https://docs.fous.com/guides/connections.md)). `connections` also lists and reads connections, names one or makes it its site's default (`update`) and disconnects one (`delete`). ## Errors A failed tool returns `isError: true` and the HTTP `error` object in `structuredContent.error`. MCP-only codes are in [Errors](https://docs.fous.com/guides/errors.md#over-mcp).