Hosted OutreachGenie MCP server
Streamable HTTP
The production endpoint is https://api.outreachgenie.ai/mcp. The server supports OAuth connections and workspace API keys. Each connection is scoped to the authorized OutreachGenie workspace. Keep API keys in an environment variable, do not commit them, and rotate them if exposed.
Create a campaign review in n8n (English guide)
What the MCP tools cover
- Verify the authorized workspace
- List and create prospect lists
- Look up, import, attach, and update prospects
- List, inspect, and create outreach flows
- Inspect the template fields available to a flow
- Prepare a runtime campaign from an approved flow and list
Connecting Claude or ChatGPT with OAuth instead of an API key? Follow the LinkedIn MCP setup.
Approval boundary
Read operations can verify access and inspect workspace state. Before creating a list, importing or updating a prospect, creating a flow, or preparing a campaign, show the intended change and ask the user to approve it.
Free saved-list audit skill
Check saved prospect lists against a saved outreach flow. The local helper reports missing personalization values, message-length problems and incomplete list captures. It does not change workspace records or launch outreach.
Try the fictional demo without an account. Python 3.10 or newer is required. Auditing your own lists also requires an OutreachGenie workspace and an MCP host that can read the selected records and run Python locally.
The package and reports are in English. The helper runs offline; during connected collection, your AI host processes the returned records. A report does not establish that a campaign is ready to send.
Get the audit skill and demo on GitHubTyped runtime API
OpenAPI 3.0.3
The official OpenAPI document describes the authenticated runtime endpoints used by the extension and MCP server. Every operation has a unique operation ID, description, typed request data, and response schema. The production server is https://api.outreachgenie.ai. Direct Runtime API requests use a workspace API key in the Authorization bearer header.
Errors an agent can recover from
API discovery routes return JSON or RFC 9457 problem details, including a stable code, a plain-language detail, and a resolution hint. The MCP endpoint returns JSON-RPC errors. Direct Runtime API requests require a workspace API key. MCP authentication failures include a Bearer challenge for OAuth clients; scope errors identify the required permission.
Choose the integration by its actual action
Zapier's native LinkedIn connector describes publishing shares and company updates. That does not establish native member-message or connection-invitation actions. Check the exact connector, action and account authorization before planning a workflow.
OutreachGenie API and MCP connections operate on the authorized workspace under their own contract. They do not grant native LinkedIn developer permissions. Keep publishing, workspace preparation and LinkedIn-side execution distinct when choosing an integration.
Connect Codex or Claude Code
Command line setup
OutreachGenie does not publish a separate general-purpose CLI. The supported command line path is the MCP command built into the agent host. Follow the official setup page for the exact command, keep the bearer token in an environment variable, then restart the agent host so it can discover the tools.
Stable resource URLs
Machine-readable discovery
Agents and registries can use the files below without signing in. They identify the authenticated service and its documentation; they do not expose a user's workspace data or API key.