Connect to the Stata you already own
stata-cli detects Stata and PyStata on macOS, Windows, and Linux. No remote data upload and no replacement statistics engine.
Panel data
Causal inference
Graphs
85 ms
Warm daemon execution
JSON
Agent-readable results
57
Built-in Stata topics
17–19
Stata versions supported
Agent-native by design
Prompt in.
Evidence out.
Give any shell-capable agent a real Stata runtime. Keep your datasets, estimation commands, stored results, matrices, and graphs in one persistent session.
Run code without losing context
A persistent daemon keeps datasets, macros, estimates, and frames alive across calls, with named sessions for parallel projects.
Return structured empirical results
Agents can inspect variables, r()/e()/s(), matrices, labels, filtered rows, help files, and exported graphs instead of scraping terminal noise.
One binary, full workflow
Everything an agent needs to work empirically.
Execute
Inline Stata, multi-line analysis, stdin, and complete .do files with reliable exit codes.
Inspect
Datasets, variable metadata, frames, labels, macros, matrices, and stored estimation results.
Interpret
Built-in references for regression, panel data, DiD, IV, RDD, matching, graphics, and community packages.
Deliver
Compact JSON responses, token-aware output, logs, and automatic PNG, SVG, or PDF graph export.
Quick start
Install.
Detect.
Run.
Stata 17 or newer and Python 3.9+ are all you need. Start the daemon once, then let your agent move at interactive speed.
$ pip install stata-cli $ stata-cli detect Stata detected $ stata-cli daemon start $ stata-cli --json run "sysuse auto, clear regress price mpg weight"