Connect UK planning data to ChatGPT, Claude, or any AI assistant

Ask real questions about UK planning applications, appeals and enforcement, answered from live council records — not the AI's general knowledge.

Quickest way to connect

Works with ChatGPT and Claude. No coding, no downloads, no API key to copy — you just sign in with your email.

ChatGPT

Open ChatGPT → Settings → Connectors → Create.
Paste this in as the server address:
```
https://ukplanningapi.co.uk/mcp
```
ChatGPT will ask you to sign in — enter your email, then click the link we send you. No password to create or remember.
Approve the connection. That's it — ask it a planning question.

Works as-is. If you also turn on Developer Mode (ChatGPT Settings → Advanced), it gets all 35 tools below instead of general search — better for detailed questions about officers, appeals or approval rates, but not needed to get started.

Claude

Open Claude → Settings → Customize → Connectors → Add custom connector (or go straight to claude.ai/settings/customize-connectors).
Paste the same address:
```
https://ukplanningapi.co.uk/mcp
```
Sign in with your email when asked. If you see boxes for Client ID or Client Secret, leave them empty — they're not needed here.

Free to try — 500 questions a month, no card required. The connection shows up as a key in your dashboard, where you can remove it at any time. Prefer building a shareable Custom GPT instead? See GPT Actions further down.

Try asking it…

"What planning applications have been submitted near NW3 6UP?"
"What's the approval rate for house extensions in Camden?"
"Is NW3 6UP in a conservation area or under an Article 4 direction?"
"Show me planning appeals decided in Leeds this year"
"What's the full planning history of this property?"

Developer setup

The rest of this page is for building an integration directly — config files, the REST API, and the full list of 35 tools — rather than using ChatGPT or Claude as-is. One API key works across REST and MCP, sharing the same monthly quota.

Transport

Streamable HTTP hosted at https://ukplanningapi.co.uk/mcp. No local install required — add the URL and your API key to your client's MCP config.

Pass your key via the X-Api-Key header on the connection. Stdio (local subprocess) transport is also supported for clients that launch a local process and inject environment variables directly — set UK_PLANNING_API_KEY in that case instead of a header.

Every request needs a credential — API key or OAuth — including tools/list. There's no anonymous session; an unauthenticated request gets 401. To browse the tool catalogue without either, read /.well-known/mcp.json instead — a free, always-open snapshot of every tool's name and description.

Claude Desktop

~/.claude/claude_desktop_config.json:

{
  "mcpServers": {
    "uk-planning": {
      "url": "https://ukplanningapi.co.uk/mcp",
      "headers": { "X-Api-Key": "YOUR_API_KEY" }
    }
  }
}

Cursor / VS Code Copilot

.cursor/mcp.json or workspace settings:

{
  "mcp": {
    "servers": {
      "uk-planning": {
        "url": "https://ukplanningapi.co.uk/mcp",
        "headers": { "X-Api-Key": "YOUR_API_KEY" }
      }
    }
  }
}

ChatGPT via Custom GPT Actions

An alternative to the ChatGPT connector in the quick-start section above.

Not MCP — a Custom GPT built from an OpenAPI schema, talking to the plain REST API with a regular API key. Useful if you're building a Custom GPT to share, or just prefer pasting a key over signing in.

In ChatGPT: Explore GPTs → Create (or open an existing GPT you own) → Configure → Create new action.

Import from URL and paste:

https://ukplanningapi.co.uk/openapi.json

Authentication → API Key. If you're offered a choice of Bearer or Custom: either works — for Custom, set the header name to X-Api-Key. Paste your API key as the value.
Save. The GPT can now call every /v1/… endpoint this API has — the same data as the MCP tools below, shaped as REST operations instead of named tools.

One API key, one quota, shared across REST, MCP and the ChatGPT/Claude connectors — nothing about using it from a Custom GPT costs differently.

Python / LangChain agent

from mcp import ClientSession
from mcp.client.streamable_http import streamablehttp_client

async with streamablehttp_client(
    "https://ukplanningapi.co.uk/mcp",
    headers={"X-Api-Key": "YOUR_API_KEY"},
) as (read, write, _):
    async with ClientSession(read, write) as session:
        await session.initialize()
        result = await session.call_tool("search_applications", {"postcode": "NW3"})

ChatGPT connector tools (2)

Tool	What an AI can ask
`search`	Search UK planning applications from one plain-English query.
`fetch`	Fetch one planning application in full by the id returned from `search`.

Core tools (9)

Tool	What an AI can ask
`search_applications`	“Show refused householder applications in NW3 in the last 6 months”
`get_application`	“Get the full detail for application 14-26/00123/FUL”
`get_site_history`	“What's the full planning history of this property?”
`get_location_constraints`	“Is NW3 6UP in a conservation area or under an Article 4 direction?”
`get_officer`	“What is officer Sarah Ahmed's refusal rate for extensions?”
`get_agent`	“What is this planning agent's approval rate in Camden?”
`list_councils`	“Which councils do you have data for?”
`get_council_stats`	“What is the approval rate for householder applications in Ealing?”
`get_changes`	“What new applications were submitted in Tower Hamlets today?”

Summary tools (11)

Tool	What an AI can ask
`get_country_stats`	“What is the approval rate in Wales?”
`get_country_trends`	“Has Wales's approval rate changed over 3 years?”
`get_county_stats`	“What is the approval rate in Lancashire?”
`get_county_trends`	“Is Lancashire approving more or fewer applications than 2 years ago?”
`get_council_summary`	“What application types get approved most in Camden?”
`get_council_trends`	“Show Camden's quarterly approval rate over time”
`get_ward_stats`	“What planning activity is there in Ealing Broadway ward?”
`get_district_stats`	“What planning activity is there in SE1?”
`get_postcode_stats`	“What planning activity has there been at NW3 6UP?”
`get_officer_history`	“Show me this officer's quarterly caseload over time”
`get_appeal_stats`	“What's the householder appeal success rate at Ealing?”

Appeals & enforcement tools (6)

Tool	What an AI can ask
`search_enforcement`	“Are there enforcement notices near SE1 9SG?”
`get_enforcement`	“What happened with enforcement notice 4821 — was it complied with?”
`search_appeals`	“Show allowed appeals in Camden this year”
`get_appeal`	“Get the inspector's decision for appeal APP/X5210/W/26/1234567”
`get_application_appeals`	“Was application 123-2026/0123/P taken to appeal?”
`get_council_fields`	“Which fields does Leeds actually publish before I query it?”

Area & directory tools (6)

Tool	What an AI can ask
`list_postcodes`	“List every postcode in Hampstead Town ward”
`list_addresses`	“What addresses are on Church Row, NW3?”
`list_properties`	“List Victorian houses in the Hampstead conservation area”
`search_officers`	“Find officer Sarah Ahmed — which councils has she worked for?”
`search_agents`	“Find agents from ‘Smith Planning Associates’”
`list_wards`	“Which wards does Camden have data for?”

Account tools (1)

Tool	What an AI can ask
`get_my_usage`	“How much of my monthly API allowance is left?”

Appeal coverage

Appeals are linked to a specific application via the Planning Inspectorate's own reference numbers and address matching — not every appeal matches. Only around 10.4% of Planning Inspectorate appeal cases link to a specific application in this database; the rest either matched to a property only or didn't match at all. get_appeal_stats and the appeal figures in get_ward_stats/get_district_stats/ get_officer_history reflect that linked subset, not every appeal filed against a council.