streamlit /streamlit

Streamlit — The agent-native framework for data apps and dashboards.

★ 46,366⑂ 4,453👁 341⑃ 11,654● 6,213▤ 132
PythonApache-2.0Created 2019-08-24streamlit.io
下载:1.67.01.67.0 ↓1.67.02026-11-05
+14
日增长
+86
周增长
46,366
stars
4,453
forks
341
watchers
11,654
prs
6,213
issues
132
releases

README

Streamlit

 Built in Python. Made for people and agents. Owned by your team.

Streamlit is the agent-native data app framework. Use it for dashboards, forecasting
tools, review-and-approval workflows, and any other app built around your data.





A Streamlit app is just Python, with no separate frontend or API layer. Streamlit is
batteries-included and takes care of the hard parts: a consistent, well-designed
interface, state, caching, sign-in, theming, and upgrades. Your team keeps only its own
logic, written in a language it already reads every day.

The No. 1 Python framework for data apps and dashboards.
More than a million repositories
on GitHub depend on it. Used by millions. Trusted by thousands of companies, from startups
to global enterprises.

 What teams build


  
    
      
    
    
      
    
    
      
    
  
  
    
      APPS THAT COMPUTE
      Forecasts, simulations, and models with what-if levers
      Pricing and demand simulators, risk models, and AI analysis over your data. The
      computation and the interface are the same Python.
      Try the live demo →
    
    
      APPS THAT ACT
      Review queues, approvals, sign-offs, and writeback
      Editable tables (st.data_editor) and forms put the decision in the same
      app as the data, including approving what an AI agent proposes.
      Try the live demo →
    
    
      ADVANCED DASHBOARDS
      KPI dashboards with drill-down and cross-filtering
      Fast over large data. Coming from a BI tool? Move your dashboards to Streamlit, and
      modernize them on the way.
      Try the live demo →
    
  


The source for all three is in streamlit/demos.
Streamlit is also great for AI and chat apps over your data (st.chat_input,
st.write_stream), ML and model tools, and internal data tools.

 Get started

bash
pip install streamlit


 Build with your coding agent

Run streamlit skills in
your project folder. It installs our expert skills for Claude Code and any agent that
reads the shared .agents skills folder, matched to your Streamlit version. So your agent
follows current best practices instead of whatever it memorized two versions ago.

Then ask, and mention Streamlit so the agent picks up the skills (in Claude Code,
/developing-with-streamlit works too):

 Build a Streamlit app that forecasts demand from sales.csv, with what-if levers for
 price and growth.

 Or write it yourself

Here's a complete app: a demand forecast that planners can adjust and submit for sign-off.
Save it as streamlit_app.py and run streamlit run streamlit_app.py:

python
import sqlite3

import numpy as np
import pandas as pd
import streamlit as st

st.title("Demand plan")


@st.cache_data
def load_actuals() - pd.DataFrame:
     Swap in your own query, e.g. st.connection("snowflake").query("...")
    months = pd.date_range("2025-01-01", periods=12, freq="MS")
    units = 1_000 + 40  np.arange(12) + np.random.default_rng(0).normal(0, 50, 12)
    return pd.DataFrame({"month": months, "act

Top issues

Latest release

1.67.0 1.67.0 2026-11-05
What's Changed
 New Features 🎉
 [feature] Update default light and dark named theme colors by @mayagbarnes in https://github.com/streamlit/streamlit/pull/17450
 [feature] Add required parameter to st.date_input by @lukasmasuch in https://github.com/streamlit/streamlit/pull/17165
 [feature] Allow a pixel width on st.dialog by @lukasmasuch in https://github.com/streamlit/streamlit/pull/17412
 [feature] Update default theme surfaces and sidebar fallback by @mayagbarnes in https://github.com/streamlit/streamlit/pull/17467
 [feature] Use theme primary for st.map default markers by @mayagbarnes in h

Snapshot: 2026-11-16T09:23:14Z · Data license Apache-2.0 · JSON