# MCP > Connect igeo.ai to AI assistants via Model Context Protocol-query prompts, mentions, and workspace data without leaving your editor. Canonical URL: https://app.igeo.ai/docs/mcp ## Overview Model Context Protocol (MCP) lets compatible AI assistants and agents talk to igeo.ai using a standard interface — so you can ask questions about your prompts, mentions, sentiment, competitors, and sources without copying exports or switching tabs all day. Think of MCP as a bridge: your AI tool stays in the editor or chat, while igeo.ai supplies the live workspace context behind the scenes. ## What is MCP? MCP is an open protocol for connecting AI applications (Claude Desktop, Cursor, Windsurf, and others) to tools and data sources. Instead of each vendor inventing a custom plugin format, MCP defines how clients discover tools, call them, and read structured resources — so integrations stay consistent and easier to maintain. igeo.ai exposes an MCP endpoint over HTTP (JSON-RPC). Connect with a workspace-scoped API key or a supported OAuth flow. Available tools include workspace analysis and supported content or configuration actions; the server enforces your workspace access and granted scopes. ## Why MCP matters for AI visibility GEO and AI visibility work is iterative: you tune prompts, watch competitors, and react to how models answer. MCP keeps that loop inside the workflow you already use — your coding agent or assistant can pull the latest igeo.ai state on demand instead of you manually reconciling screenshots or CSVs. - Faster decisions — Compare “this week vs last week” or “this topic vs that topic” with natural language instead of rebuilding the same filter sets by hand. - Shared context — The same workspace rules (membership, roles and granted scopes) apply through MCP, so answers stay aligned with what the dashboard would show for that workspace. - Automation-ready — Agents can chain steps (fetch prompts → summarize weak spots → draft follow-up actions) using structured tool results. ## How MCP makes igeo.ai easier to use Without MCP, jumping between igeo.ai and an AI assistant means re-explaining filters, pasting tables, and risking stale numbers. With MCP connected, you keep one thread of conversation and let the client fetch fresh workspace data when it needs it. ## Examples of what you might ask “Run prompt_work_priorities and tell me which prompt to maintain and which to improve this week.” “Which tracked prompts had the lowest visibility in the last seven days?” “Summarize sentiment for our brand vs the top two competitors this month.” “What allowlisted read routes exist for this workspace?” (discovery via MCP tools/resources) Exact tool names and capabilities evolve with the product; your client will list what the connected server exposes after setup. ## Copy-paste prompts (weekly planning) Paste one of these into Claude, Cursor, or ChatGPT after connecting igeo.ai MCP. Or fetch the full JSON playbook via MCP resources/read on igeo://mcp/playbooks/prompt-prioritization. ## Weekly planning (recommended) You are connected to igeo.ai via MCP. 1) Call the tool prompt_work_priorities with recentDays=30 (visibility uses the last 7 days by default). 2) Summarize: total prompts, how many had post activity in 30 days, visibilityRangeDays, and thresholds (maintenance ≥50, improvement ≤49 by default). 3) Present MAINTENANCE (maintenancePick) and IMPROVEMENT (improvementPick) with rationale and one concrete next action each. 4) If a pick is null, use the best alternate from alternates. Use only MCP tool results — do not guess. ## Audit recent work Using igeo.ai MCP: 1) Call prompt_work_priorities with recentDays=14. 2) List recentlyWorked sorted by lastPostAt (newest first) with topic, post count, and avgVisibility. 3) Flag high-visibility prompts with only one post in the window. ## Deep dive on improvement pick Using igeo.ai MCP: 1) Call prompt_work_priorities with recentDays=30. 2) Use improvementPick.promptId with prompt_responses (query.range=30). 3) Explain why visibility is low and propose 3 content actions. ## How to connect MCP and API keys are available to active workspace members. Select the correct workspace before creating a key. If the setup page is unavailable, check your membership and account status with a workspace admin. 1. Sign in to igeo.ai, select the workspace and open MCP & API keys. 2. Create a key — The app generates a server URL and client configuration (JSON) you can paste into your MCP client. Some clients accept a single URL with token query parameters; others want headers (Authorization, X-Workspace-Id) — use the snippet that matches your tool. 3. Point the client at the igeo.ai MCP server — Requests go to your environment's API base with the /mcp path (for example the same host your app already uses for VITE_API_URL, plus /mcp). 4. Start a session — Initialize the MCP connection in the client, run tools/list (or the equivalent in the UI), then ask questions that trigger tool calls. Use only the tools and operations permitted for your credentials. The api_get tool is read-only; supported write tools also require the appropriate role and scopes. ## Connect with supported OAuth clients When a compatible client starts an OAuth connection, follow its sign-in flow, select the intended workspace and review the permissions requested. Approve only the workspace and scopes required for that integration, then return to the client and verify the available tools. If a Claude connection link is invalid or expired, restart the connection from Claude. API keys are shown in full only when created. Store the secret securely, use the configuration supplied by the app and revoke a key when it is no longer needed. Client cards with an active-key status do not independently verify that every external client is connected. For workspace switching, billing, and team roles, see Manage your project. ## Data use and license The same data-use license that applies to the Public API applies to MCP. You may use data lawfully retrieved through MCP for any lawful purpose of your choosing. See Public API — Data use and license. ## Related Public API Build direct REST integrations with the same workspace API keys. Manage your project Workspace settings, team, billing, and where account-level tools live. Quickstart Guide Get prompts and workspaces set up so MCP-backed questions return meaningful data. Sidebar navigation Find Account, analytics, and monitoring areas in the app. ## Video walkthroughs - [Connect igeo.ai MCP to Claude](https://www.youtube.com/watch?v=YcLvqjPPEMc)