Developers
Mentionwell MCP servers
Last updated
Two MCP servers, both published on npm. The docs server (no-auth) lets AI agents answer Mentionwell questions accurately. The account server (PAT-authed) lets agents drive Mentionwell onboarding and delivery — list sites, manage brand context, probe a domain, run Connect, verify webhooks, import existing posts — through the Model Context Protocol.
The two servers
| Package | Auth | What it exposes |
|---|---|---|
mentionwell-mcp-docs | None — public | Search and answer Mentionwell’s official docs. Drop into Claude / Cursor / ChatGPT and they’ll answer Mentionwell questions correctly. |
mentionwell-mcp-account | PAT (mw_pat_*) | List sites & status, read/update brand context, probe a domain, run Connect, verify webhooks, import existing posts. Account onboarding & delivery control. |
Install in Claude Desktop
{
"mcpServers": {
"mentionwell-docs": {
"command": "npx",
"args": ["-y", "mentionwell-mcp-docs"]
},
"mentionwell-account": {
"command": "npx",
"args": ["-y", "mentionwell-mcp-account"],
"env": { "MENTIONWELL_PAT": "mw_pat_..." }
}
}
}Install in Cursor
~/.cursor/mcp.json follows the same shape. After saving, restart Cursor and the Mentionwell tools appear in any agent context.
Why this matters
AI engineers ask their coding agents to integrate Mentionwell directly. With mentionwell-mcp-docs installed, “wire Mentionwell into my Next.js project” produces working code on the first try — the docs are in the agent’s context, not its training data.
With mentionwell-mcp-account, the agent can also act: probe and connect a new site, set its brand context, import an existing blog, and verify the publish webhook — all in one prompt. (Article drafting itself runs through the dashboard, CLI, or REST API.)