Aperiodic
Metrics
MarketsLive
DocsPricing
Get Started

Start with your agent

Find the cheapest venue to trade

What a market order costs on each venue, and when. Your agent installs the client, tries the free preview, then does this.
Start in Claude DesktopStart in Claude Code (terminal)Start in Codex

Opening Claude Desktop with this prompt typed in. Nothing opened? It may not be installed here: copy the prompt into any agent.

paste into your agent
Read https://aperiodic.io/start.md and follow it. Then, for the HUMA perps on every venue over the last 30 days, compare what a $100k market order costs (`l1_liquidity`, `slippage`, `impact`), and tell me where and at what hours it is cheapest.

What the line does

It sends your agent to one short Markdown file, /start.md, and asks it to follow it. You can read the file before you paste the line. On a first run it tells your agent to:

  1. 1

    Install the Python client from PyPI with pip install "aperiodic[polars]", in a virtualenv if your project has none. If the Aperiodic MCP tools are already connected in your app, skip this and fetch the preview through them.

  2. 2

    Fetch the free preview, which needs no key: taker order flow (flow) for BTC on Binance in 5m bars, 2025-05-01 to 2025-05-31.

  3. 3

    Reply in a few lines: the five bars with the biggest net flow as a small table, how often buyers led, three questions it could answer with a key, and where to create one.

  4. 4

    If you typed a question after the line, work on that instead, and ask once for a key if it needs data beyond the preview.

  5. 5

    Ask you before keeping anything: with your yes, save the file as a skill and add Aperiodic as an MCP server.

After that, it reads your key from APERIODIC_API_KEY, or asks you for it once.

If your agent can't reach aperiodic.io

Agents that run in a cloud sandbox can only reach the domains their app allows. If the first run stops at the fetch or the install, allow these:

aperiodic.io
The file itself, and the API the client calls.
trade-metrics.aperiodic.io
The preview's Parquet file. *.aperiodic.io covers the files of every dataset.
pypi.orgfiles.pythonhosted.org
Where pip installs the client from. Sandboxes that allow package managers usually allow these already.

Where to allow them:

Claude on the web or desktop
Code execution's network access, under Settings > Capabilities. On Team and Enterprise plans an owner sets the allowed domains under Organization settings > Capabilities.
Claude Code on the web
The environment's network access: add them as allowed domains.
Codex cloud
The environment's agent internet access, off by default: turn it on and add them to its allowlist.
On your own computer
Claude Code, Codex CLI or Cursor may ask before fetching a page or running pip. Each request is one of the steps above.

Every prompt we link to

Buttons across the site, from the catalog and the API docs to notebooks, the guide and Pre-flight, open your agent with one of these. Each reads a skill file and then does one task. The underlined words are what the page you came from fills in.

Aperiodic

Built from hundreds of terabytes of raw data and distilled into point-in-time metrics you can pull as parquet files. Plus the raw trades, quotes and derivatives data behind them.

Use Cases
  • Improve your backtest
  • Find new alpha
  • Execute smarter
Indices
  • All indices
  • Aperiodic Cost Index
  • Aperiodic Depth Index
  • Aperiodic Impact Index
Metrics
  • Metrics library
  • Data Catalog
  • Live markets dashboard
  • Global Liquidity
  • Unusual activity
  • Order Flow
  • L1 & L2
  • Derivatives
  • RWA
  • Raw Data
Developers
  • API Docs
  • MCP Server
  • Notebooks
  • For AI Agents
  • Atlas Security Master
  • Pre-flight
  • Changelog
  • Roadmap
  • llms.txt
Learn
  • Replays
  • Microstructure Guide
  • Architecture
  • Research Blog
  • FAQ
Compare
  • All vendors
  • vs. Tardis
  • vs. CoinGlass
  • vs. Kaiko
  • vs. Amberdata
Company
  • Team
  • Pricing
  • Contact
  • Book a call
  • Terms of Service
  • Privacy Policy
  • Licensing
  • SLA

© Copyright 2026 Aperiodic. All Rights Reserved.

Aperiodic Limited · Registered in Ireland · Company No. 815273 ·
136 Capel Street, Dublin D01 T2C9, Ireland