MCP servers

What this is for. MCP (Model Context Protocol) is how EYAS attaches external toolboxes: a filesystem server, a SaaS MCP, a local npx process. Tools discovered here become assignable like builtins. This is not a chat channel and not a Connection inventory row — though you can also register an MCP server as a Connection for health tracking.

Route: /mcp-settings. Title: MCP Servers. Subtitle: Extend EYAS with external tools, resources, and prompts via Model Context Protocol. Tabs: Active · Catalog.

When to use it

An agent needs tools that EYAS does not ship (a vendor MCP, a local filesystem server).
You want a one-click catalog install (API key) instead of typing a command.
Grok/Kimi CLI sessions should see the same ToolExecutor tools as in-process agents.
A server is disconnected and you need Test / discovered tool counts.

Typical workflow

Open MCP Servers (/mcp-settings).
Browse Catalog. Filter by category. Ready to use / One-Click Install (API Key Required) / Third-Party (Manual Setup).
Install (fill env keys if prompted) or Manual → Add MCP Server (name, transport, command or URL).
On Active, confirm status connected, run Test, inspect discovered tools / resources / prompts.
Assign those tool ids on the agent Configuration tab. See Tools.

Features

Header shows N/M connected. Catalog entries carry a license badge (MIT-compatible / copyleft / proprietary / unknown) — copyleft and proprietary still run as a separate process; EYAS remains MIT.

You can also register an MCP server as a Connection inventory row (type MCP server) for health tracking next to Odoo/GitHub/etc.

Magnific, Higgsfield, and fal connect under Media; the agent uses five media_* tools instead of their raw MCP catalogues.

Agent Browser (Vercel, Apache-2.0) is a Browser catalog row: agent-browser mcp --tools core,state. Install the CLI first (EYAS_AGENT_BROWSER_BIN or PATH). Never --tools all (includes chat). See Browser Use.

Chrome DevTools MCP (Google, Apache-2.0) is a DevTools catalog row: npx -y chrome-devtools-mcp@latest --isolated with telemetry off and --categoryExperimentalWebmcp=true. Coding/debug only (console, network, Lighthouse, WebMCP) — not form-filling. Tools arrive as mcp_chrome-devtools_*. WebMCP tools (list_webmcp_tools / execute_webmcp_tool) only if the sidecar advertises them; otherwise they are not invented. --autoConnect and the daily Chrome profile are refused. See Browser Use.

Fields and controls

Active servers

Control	Meaning
disabled	Server exists but is not enabled
Test	Probe the connection
Connect	OAuth servers: starts the browser sign-in (`POST …/oauth/start` → redirect)
Managed by Settings → Media	Shown when `ownedBy` is `media`
N tools / N resources / N prompts	Discovered catalogue
Connection OK / Test failed	Last test
Edit / delete	Change command, URL, API key

Add / edit dialog

Field	Meaning
Name	Display id
Transport	stdio (local process) · HTTP (remote) · SSE (Streamable HTTP) — the `sse` transport is Streamable HTTP; do not append `/sse`. EYAS handles the session header.
Command / Arguments	stdio process (`npx` + space-separated args)
URL	HTTP / SSE endpoint (no `/sse` suffix)
Authentication	None · Bearer (API key) · OAuth (browser)

Catalog

Control	Meaning
Category filter	All (N) plus per-category
Install / Installed	One-click or already present
Setup guide	Expand vendor instructions
Env dialog	Required keys before Install & Connect
License notice	Licensed under … Runs as a separate process — EYAS remains MIT.

Empty active: No MCP servers configured — Browse Catalog.

CLI MCP tool parity (Grok / Kimi)

API and in-process providers already share EYAS tools. For host CLI providers:

Provider	Behaviour
Claude Code	In-process MCP (existing)
Grok CLI / Kimi Code CLI	Stdio MCP server + loopback bridge (`/api/v1/internal/cli-mcp/*`) with short-lived secrets; ACP `session/new` receives `mcpServers` so the CLI host can call the same ToolExecutor tools

Result: coding CLIs and the web agent path see a consistent tool surface instead of inventing parallel integrations.