Skip to main content
Glama
London E-Bike Parking Bays

London E-Bike Parking Bays

by jhammant

London e-bike parking bays (remote MCP server)

Ask Claude "where can I park the bike?" while cycling in London. One tool, nearest_bays(lat, lng), returns the three closest council dockless bike hire bays as short spoken lines ("Pancras Road, 80 metres, north-east"), with coordinates in the structured result.

  • MCP over Streamable HTTP at /mcp (stateless, no auth), plus GET /health and GET / (what it is + attribution).

  • Bundled data: Camden and the City of London (Open Government Licence). Sources: SOURCES.md.

  • Optionally serves a bigger bay index you mount yourself (see below).

Run it yourself

uv venv .venv -p 3.12 && uv pip install -p .venv -r requirements.txt pytest
python3 -I scripts/build.py            # data/raw -> data/bays.sqlite (SQLite R*Tree)
PORT=18765 .venv/bin/python -m app.server
.venv/bin/python scripts/smoke.py http://127.0.0.1:18765/mcp
.venv/bin/python -m pytest -q tests

Or with Docker (the image builds the OGL index itself):

docker build -t lime-bays .
docker run -p 8000:8000 lime-bays      # -> http://localhost:8000/mcp

To use it from claude.ai, put it behind HTTPS on a host you control and add https://<your-host>/mcp under Settings, Connectors, Add custom connector. Ship only committed files (for example git archive HEAD) so local venvs or data files never reach the build context.

Rate limit: a sliding one-minute window per client IP (in memory; RATE_LIMIT_PER_MIN, 300 in the image). The client IP is taken TRUSTED_HOPS (default 1) entries from the right of X-Forwarded-For; if your proxy forwards only its own address, every caller shares one bucket.

Related MCP server: gbfs-mcp

Licence

The code is MIT (see LICENSE). The bay data in data/raw/ is not: it contains public sector information licensed under the Open Government Licence v3.0 (London Borough of Camden; City of London Corporation). Sources and attribution are in SOURCES.md.

A bigger index of your own

You can mount a second SQLite index with the same bays + bays_rtree tables (extra columns status and approximate are honoured: only Active bays are returned, approximate ones are flagged in speech). Keep it out of this repo and out of the image; mount it read-only, for example -v /srv/bays:/private:ro.

Variable

Effect

PRIVATE_DB_PATH

Path to the mounted index (the image sets /private/bays_private.sqlite).

PUBLIC_DATASET

ogl (default): the main /mcp serves only the bundled OGL data. full: the main /mcp serves the mounted index, and its tool description says which boroughs it covers. If the file is missing it falls back to ogl.

LIMEBAYS_PRIVATE_TOKEN

At least 32 random URL-safe characters: also serves the mounted index at /p/<token>/mcp (constant-time token check; any other /p/... is a plain 404; access logs show /p/***).

Only use full for data you are allowed to serve. When it is on, / keeps the OGL attribution for the Camden and City of London records.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      A
      quality
      C
      maintenance
      Enables AI assistants and developers to search 1,000+ public bikeshare systems and fetch live station, free-floating vehicle, alert, and vehicle-type data through the General Bikeshare Feed Specification.
      6
      1
      MIT
    • A
      license
      Not graded
      quality
      C
      maintenance
      Provides keyless, real-time access to London transport data, letting agents check line service status, search stops and stations, fetch live predicted arrivals, plan journeys across modes and postcodes, see Santander cycle-hire dock availability, and list current disruptions. It can be used as a hosted remote MCP endpoint, a local stdio server, or plain HTTP calls.
      292 npm
      MIT
    • A
      license
      Not graded
      quality
      C
      maintenance
      Turns a local business's menu, hours, and booking rules from one JSON config into voice-callable tools, so an Alexa+ assistant can answer questions, quote and place pickup orders, check order status, and book appointments with customer confirmation. Works for restaurants, barbershops, salons, and clinics as a self-hosted Streamable HTTP server with a live owner dashboard.
      MIT