Localnord MCP¶

Remote MCP server (Streamable HTTP, OAuth 2.1 with PKCE) - connects Claude or any MCP client to a local marketing workspace: Google Business Profile listings, reviews, stats, posts and photos across an organization's locations, with audits and rank scans, behind granular per-scope permissions.

Server type: Remote (Streamable HTTP)
Auth: OAuth 2.1 with PKCE (Sign in with Localnord; scopes chosen at consent)
Endpoint: https://app.localnord.io/api/mcp/mcp
Tools: listings, reviews, stats, audits, drafts, posts, photos
Pricing: Localnord plans (paid actions consume credits)
Built by: Localnord
Listing: mcpservers.org (Oct 2026)

Why This Matters for Operators¶

For a local business, or an agency running many of them, the Google Business Profile is the front door: hours, services, photos, posts and review replies are what customers see before the website. Localnord is the management platform for that surface, and its MCP server turns the work into chat: pull the listing and review state, draft replies and posts, run local SEO audits and rank scans, then publish what the owner approves.

The design keeps the risky parts safe: drafts are created first and nothing is published without approval, write actions are split into their own scopes, paid actions are confirmed before running, and every connection is scoped to a single organization and revocable from Settings.

Permissions (Scopes)¶

Scope	Grants
`mcp:read`	Read-only: listings, reviews, stats; run audits
`mcp:write:drafts`	Create post and reply drafts (nothing is published without approval)
`mcp:write:reviews`	Post replies to reviews directly on Google
`mcp:write:content`	Publish posts; upload and delete photos
`mcp:write:listing`	Edit hours, description, services and attributes of the listing
`mcp:write:paid`	Full audits, rank scans and websites (consume credits)
`mcp:admin`	Everything above plus sensitive actions (SMS, webhooks, portal)

Installation¶

claude mcp add --transport http localnord https://app.localnord.io/api/mcp/mcp

Or declare the remote server in any MCP client (Cursor, VS Code, others); the OAuth flow starts on first connection. The consent screen lists the scopes, hierarchical and granular, so you only grant what you need.

FAQ¶

What can the agent do without publishing anything?¶

Read listings, reviews and stats, run audits, and create drafts of posts and review replies. Publishing requires the write scopes.

Is one connection enough for multiple locations?¶

Yes, within your organization. Each connection is scoped to it, no other agency's data is reachable, and demo organizations cannot connect the server.

Which actions are rate limited?¶

Sensitive or paid actions (full audits, rank scans, SMS, webhooks) are rate-limited against abuse and confirmed before running; paid actions consume credits.

How do I disconnect or revoke access?¶

Settings, API and MCP. Every connection is individually revocable, and permissions are chosen per connection at sign-in.

Integration with CorpusIQ¶

CorpusIQ is the connector layer that lets AI assistants read the systems a business already runs on - accounting, payments, CRM, email, files and analytics - and answer with the source named. It is read-only by design: it retrieves data and never writes to, or acts in, a connected system, which is what makes it safe to point at live accounts.

This page is part of the CorpusIQ directory, the reference we maintain for the MCP server and agent-skill ecosystem. Run alongside CorpusIQ, a third-party server like this one can be combined with a company's own business data in a single answer.

See what CorpusIQ connects