The problem

You have 30 Pandera schemas spread across a data project. New team members ask: "Which columns does OrdersSchema have? Is amount nullable? What checks run on user_id?"

The answer is buried in code. There's no docs page, no searchable index — just grep and hope.

The solution

pip install pandera-ui
pandera-ui /path/to/myproject

pandera-ui scans your project, discovers every DataFrameSchema and DataFrameModel, and opens a Swagger-like UI at http://localhost:8765.

Features

Quick start

# Install
pip install pandera-ui

# Scan current directory and open the UI
pandera-ui .

# Scan a specific project on a custom port
pandera-ui /path/to/myproject --port 9000

Terminal output (with pandera-ui[rich]):

Found 4 schema(s).

 Schema    Type             File                   Columns
 orders    DataFrameSchema  dataframe_schemas.py   5
 products  DataFrameSchema  dataframe_schemas.py   4
 users     DataFrameModel   schema_models.py       4
 events    DataFrameModel   schema_models.py       5

UI ready at http://127.0.0.1:8765

Open http://localhost:8765 in the browser — searchable sidebar, column table with dtypes and checks, AST/runtime badge, dark/light theme.

Installation

pip install pandera-ui            # core
pip install pandera-ui[rich]      # + spinner and summary table
pip install pandera-ui[watch]     # + --watch live-reload
pip install pandera-ui[rich,watch]  # everything

With uv:

uv add pandera-ui
uv add pandera-ui[rich,watch]

Requires Python 3.10+.

Docker

docker run --rm \
  -v /path/to/myproject:/project:ro \
  -p 8765:8765 \
  ghcr.io/darius-krsk/pandera-ui:latest

All CLI options

# Export schema docs (no server started)
pandera-ui . --export markdown > schemas.md
pandera-ui . --export html    > schemas.html

# Check documentation coverage
pandera-ui . --coverage

# Live reload on .py changes
pip install pandera-ui[watch]
pandera-ui . --watch

# Export raw JSON for CI / tooling
pandera-ui . --json > schemas.json

Full CLI reference:

Usage: pandera-ui [OPTIONS] [PROJECT_PATH]

Arguments:
  [PROJECT_PATH]  Project root to scan  [default: .]

Options:
  -p, --port INTEGER           Port for the UI server  [default: 8765]
  --host TEXT                  Host to bind  [default: 127.0.0.1]
  --json                       Print JSON to stdout, no server
  --export [markdown|html]     Export docs to stdout, no server
  --coverage                   Print coverage stats and exit
  -w, --watch                  Auto-reload schemas on .py changes
  --help                       Show this message and exit.

What gets extracted

Per column: name, dtype, nullable, required, checks (with parameters), title, description.

Per schema: name, coerce, title, description, index, source file, variable/class name.

Python API

from pandera_ui import scan_project

schemas = scan_project("./myproject")
for schema in schemas:
    print(schema.name, [c.name for c in schema.columns])

scan_project returns a list of SchemaMetadata Pydantic models — serialize with .model_dump().

# Compute coverage programmatically
from pandera_ui._coverage import compute_coverage, format_coverage
report = compute_coverage(schemas)
print(format_coverage(report))

# Export to Markdown
from pandera_ui._export import to_markdown
print(to_markdown(schemas))

Architecture

pandera_ui/
  scanner.py            # discovery: walks project, dispatches per file
  _extract_runtime.py   # pass 1: dynamic import + introspection
  _extract_ast.py       # pass 2: static AST parse (fallback)
  models.py             # Pydantic models: SchemaMetadata, ColumnMetadata …
  server.py             # FastAPI: GET /api/schemas, GET /api/coverage, GET /
  cli.py                # Typer CLI entry point
  _export.py            # Markdown and HTML renderers
  _coverage.py          # Documentation coverage calculator
  _console.py           # optional rich output (spinner, summary table)
frontend/
  index.html            # single-page UI (vanilla JS, no build step)

Development

git clone https://github.com/Darius1223/pandera-ui
cd pandera-ui
uv sync

make lint        # ruff check
make type        # mypy
make test        # unit tests
make test-cov    # unit tests + coverage report
make run         # start UI against test fixtures

See CONTRIBUTING.md for PR guidelines.

Changelog

See CHANGELOG.md.

License

MIT © 2025 Ildar