Ping·claw · v1.0 · iOS + Android

Location context
for any AI agent.

A quiet iOS and Android app that shares your location with the AI tools you already use — Claude Code, Cursor, OpenClaw, or anything that speaks MCP. Open source. Self-hostable. Your location is never persisted.

Phone-first sign-in · no SMS 24-hour TTL, Redis only MIT licensed
location live
39.9043° N −75.3505° W
acc ± 7m src gps age 4s ttl 24h
§ 01 / AGENTS

Works with the agent you already run.

PingClaw is a location source, not an agent companion. Any agent that speaks MCP or accepts a webhook can consume your location — configuration snippets for all of them are one tap away in the dashboard.

GPT · One-tap ChatGPT Open the PingClaw GPT. The easiest path in.
MCP · Pull Claude Code Ask "where am I" — gets coords in the loop.
MCP · Pull Claude Desktop Via mcp-remote. Paste the snippet, done.
MCP · Pull Cursor Add to MCP settings. Works in chat and agent mode.
MCP · Pull Zed Context servers. One line in settings.
MCP · Pull VS Code MCP extension support. Shared with Copilot.
Push · Native OpenClaw Direct gateway delivery. No polling needed.
Webhook · Push Anything else POST with a bearer secret. Roll your own agent.
$ terminal
# register a webhook with your API key curl -X PUT https://pingclaw.me/pingclaw/webhook \ -H "Authorization: Bearer ak_…" \ -H "Content-Type: application/json" \ -d '{"url":"https://your-receiver/location","secret":"your-secret"}'
§ 02 / SETUP

Three steps. The app comes first.

Install the app, sign in on the phone, then open the web dashboard with a code the app hands you. No phone number, no SMS, no account to recover.

STEP 01

Install the app

iOS or Android. Sign in with Apple or Google inside the app, grant background location, and you're done with the phone.

STEP 02

Open the dashboard

From the app, generate a one-time code and paste it below. That's how the site knows it's you — no separate login.

STEP 03

Connect your agent

In the dashboard, mint an API key and copy a one-line config for your client. Or register a webhook for anything else.

§ 03 / IN USE

A quiet utility,
not an app you stare at.

PingClaw runs in the background. When your AI assistant needs your location, it gets the coordinates — accurate, current, scoped to the one API key that asked.

Coordinates only. Current only. Nothing is ever written to Postgres.

Coordinates lat · lon · accuracy
Cache lifetime 24h · Redis only
Persisted history none
Map rendered none
Notifications none
Scope one key, one user
§ 04 / OPENCLAW

Native push for OpenClaw users.

For OpenClaw, PingClaw goes a step further and pushes each location update directly to your gateway. Your agent gets your position as context automatically — no polling, no manual queries.

Two ways in.

Install the PingClaw skill to let your agent fetch your location on demand, or configure gateway push delivery from the dashboard for real-time updates.

Push mode supports two actions: Wake, which injects location as context silently, and Agent, which triggers a full agent run on every update.

push flow
phone pingclaw gateway
pull flow (skill)
agent skill pingclaw
§ 05 / SELF-HOSTING

Run your own server.

PingClaw is open source and self-hostable. One binary, no dependencies — SQLite, in-memory cache. No Redis, no Postgres, no outbound calls you didn't configure yourself.

Install the app as usual. On the sign-in screen, pick Self-Hosted Server, enter your server URL, and paste the pairing token your server printed on first run. No Apple or Google account needed — your location never leaves the machine you chose.

See the server repo for details.

$ terminal
# install go install github.com/pingclaw-me/pingclaw-server/cmd/pingclaw-server@latest # run — your location never leaves this machine pingclaw-server --local # listening on :8080 # dashboard at http://localhost:8080 # pairing token printed here — paste into the app

Give your agent a sense of place.

Install the app, pair your agent, and the next thing it answers about "near me" will actually mean something.