Glassnode has published a workflow demonstrating how traders can combine AI coding agents with its command-line interface to automate on-chain research—turning plain English questions into statistical analysis and charts within minutes.

The blockchain analytics firm showcased the approach on April 1, walking through a real example: testing whether extreme BTC exchange inflows predict short-term price drawdowns. The entire analysis—from data retrieval to visualization—required just two natural-language prompts.

How It Actually Works

The workflow hinges on Glassnode’s CLI tool, which lets AI agents like Claude Code, ChatGPT’s Codex, or Cursor autonomously discover available metrics, fetch data, and execute Python analysis without manual API configuration.

In Glassnode’s demonstration, a user submitted this prompt: “Download BTC daily exchange inflows and closing price for the last year. Analyze whether inflow spikes predict drawdowns in the following 7 days.”

The AI agent then handled everything behind the scenes—running gn metric list to find the right data paths, downloading CSVs for exchange inflows and price data, writing statistical analysis code, and returning formatted results.

What the Test Found

The sample analysis flagged 10 “spike days” where BTC exchange inflows exceeded two standard deviations above the mean. Those days showed roughly 1.9 percentage points more drawdown over the following week compared to normal periods.

Glassnode was quick to note the limitations. With only 10 spike events concentrated in two volatile windows, the signal is “suggestive rather than statistically robust.” A proper backtest would need to control for volatility regimes, avoid overlapping measurement windows, and validate out-of-sample.

Still, the point wasn’t to prove the hypothesis—it was to show how quickly traders can test ideas.

Practical Applications

The firm suggested several starter prompts for users exploring the workflow:

• “Download ETH staking deposits for the last 6 months and plot the trend”
• “Compare BTC and ETH exchange netflows over the last 90 days”
• “Find which metric has the highest correlation with BTC 30-day returns”

This follows Glassnode’s broader push to make on-chain data more accessible. The company published a guide on building crypto research strategies using its platform just two weeks prior on March 17.

Access Requirements

There’s a catch for retail users: the CLI requires an API key available only to Glassnode Professional subscribers. The tool supports integration with most major AI coding assistants through an optional skill configuration.

For institutional desks and serious researchers already paying for Glassnode access, though, the workflow could meaningfully compress the time between hypothesis and initial results—particularly for screening ideas before committing to full backtests.

Image source: Shutterstock