Your AI agent trades. You stay in control.

OpenTicker is a self-hosted trading platform built for AI agents. It turns your broker into tools any agent can call, puts every order through risk checks and a paper sandbox, and shows you each move as it happens.

Runs on your machinePaper-firstWorks with any MCP client
Plugs into how you already work
Any MCP clientREST APIPython scriptsWebhook alertsSlackEmailBroker adaptersWeb dashboard

How an order moves through OpenTicker.

  1. 01
    Your agent asks

    It calls a typed tool, the same way it reads a file. No scraping, no copy-paste.

  2. 02
    Risk checks run first

    Stops, targets, capital caps and the market calendar are checked before anything moves.

  3. 03
    The sandbox fills it

    Live prices, simulated fills. Waiting orders fill when the price gets there. Nothing reaches your broker.

  4. 04
    It goes on the record

    Every order, fill and change says who did it: you, your agent, a strategy, an alert or a script.

  5. 05
    You hear about it

    Fills and risk events reach you on Slack or email, and you can step in from the desk.

New orderFrom agent
“Buy one lot when the morning high breaks, stop below the low.”
place_orderSL-M · paper
Risk checksBefore entry
Market is open
Stop loss is set
Within your capital cap
Daily loss limit not hit
Kill switch is off
Live prices
Sandbox
Your broker
Order statePaper
Waiting for price
Filled
ActivityAudit log
AGENTPlaced a stop orderpaper
STRATEGYMoved stop to entrytrailing rule
ALERTWebhook signal acceptedsignal strategy
SCRIPTStarted its daily runscoped key
YOUClosed a positionfrom the desk
Order filled · paperSlack · just now
Leg stop hit, position closedEmail · just now
Session expired, log in againSlack · just now

Everything a trading desk needs. Built for agents.

Agents don't need another chat window. They need tools, limits and a record. OpenTicker gives them all three, and gives you the controls.

  • Agent-native tools

    Every brokerage action is a typed MCP tool with described inputs: quotes, candles, option chains, orders, positions.

  • Paper sandbox

    Orders fill against live prices in a local sandbox. Market, limit, stop-loss and stop-market orders all behave like the real thing.

  • Risk guardrails

    Per-leg stops, targets and trailing stops, strategy-wide limits, a daily loss limit, capital caps and a kill switch.

  • Strategies that run themselves

    Multi-leg options strategies chosen relative to the market, entered on schedule, skipping holidays, squared off before the close.

  • Signal strategies

    Point a charting alert at a strategy's own webhook. Alerts enter and exit positions, with an IP allowlist and rate limits.

  • Your own scripts

    Upload Python and schedule it. Each run gets a key that reaches prices, orders and positions only, never your broker secrets.

  • Full audit trail

    Every action is recorded with who triggered it. Results are shown after charges, so you see what a strategy really made.

  • Agent reviews

    Your agent reviews a strategy's runs on a schedule and writes a verdict: keep it, change one thing, or retire it.

  • Broker adapters

    Brokers plug in behind one interface, so strategies, tools and the desk work the same whichever broker you connect.

A live desk for you.

Your agents work in the background. Every board updates as they act, so you always know what is open, what is waiting and who did it.

TodayLive
Activity
STRATEGYStop hit, leg closedper-leg stop
AGENTReview writtenchange one thing
SCRIPTRun startedscoped key
ALERTSignal acceptedwebhook
STRATEGYLegs filledpaper
AGENTScheduled a strategyenters at the open
Click any board

Positions, orders, strategies, the option chain and your agents, all live.

⌘K from anywhere

Place a paper order, close a position or pause a strategy without leaving the page.

What needs you, first

An expired session or a stopped strategy sits at the top with the button that fixes it.

Prices come in. Orders don't go out.

OpenTicker runs on your machine with your own broker keys. Your agent gets live market data; your account never gets a surprise.

  • Keys encrypted at rest with a key made on your machine.
  • Listens on your machine only unless you choose otherwise.
  • Your data stays yours: history, orders and logs in a local database.
YOUR MACHINEYour agentOpenTickerYour dataBrokerPRICESNO ORDERS

Questions

About safety, setup and what it works with.

Does it place real orders?

No. Orders are paper trades in a local sandbox that fills at live prices. Nothing is sent to your broker.

Which brokers can I connect?

Brokers plug in through adapters behind one interface. The supported list is in the README, and new adapters are added on demand.

Which AI agents work with it?

Any agent that speaks MCP. Scripts and other tools can use the same actions over a REST API.

Where does my data go?

Nowhere. History, orders, logs and encrypted broker sessions stay in a local folder on your machine.

Can strategies run while I'm away?

Yes. The server enters scheduled strategies, watches stops and targets on live prices, and closes intraday positions before the session ends. If prices stop arriving, it stops the run safely.

What do I need to run it?

Python, the uv package manager, and API access from your broker. The quickstart takes you from clone to first paper trade.

Give your agent a desk.

Clone the repo, add your broker keys, and connect your agent. Your first paper trade is a sentence away.

$ git clone https://github.com/kushalBanda/openticker.git