🔎 Check Dependencies

check-dependencies scans 🐍 Python imports and compares them with the dependencies declared in pyproject.toml.

It can be used locally or in CI/CD pipelines to find dependencies that are missing from the project configuration or declared but not actually used.

It supports PEP 621, Poetry (v1.2+), Hatch, and legacy tool.uv dependency configuration.

This is a pure-Python package with no runtime dependencies on Python 3.11+ (toml is only required on older Python versions).

The project also ships a secondary CLI, dependency-writer, which writes package-to-import mappings to a TOML config file. This is useful for creating or updating [tool.check-dependencies.provides] entries.

📦 Installation

Install with uv:

uv tool install check-dependencies
check-dependencies

Install with pipx:

pipx install check-dependencies
check-dependencies

Run without installing:

uvx check-dependencies
pipx run check-dependencies

GitHub Action

Use this repository as a reusable GitHub Action in third-party workflows.

The action installs this package in an isolated virtual environment and runs python -m check_dependencies with the configured inputs.

Usage

name: Check dependencies

on:
  pull_request:
  push:

jobs:
  check-dependencies:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v6
      - uses: actions/setup-python@v6
        with:
          python-version: "3.14"  # any Python version supported by check-dependencies
      - uses: schollm/check-dependencies@v1
        with:
          output-format: github
          file-names: |
            src/
          include-dev: "false"
          known-extra: ""
          known-missing: ""
          provides: ""
          includes: ""
          provides-from-venv: ""

Inputs

• file-names (required): newline-separated paths to files/directories
• output-format: output format (concise, full, or github); default github
• known-extra: comma-separated package list
• known-missing: comma-separated module list
• provides: comma-separated PACKAGE=MODULE mappings
• include-dev: true or false
• includes: newline-separated list of additional config files
• provides-from-venv: path to venv Python executable

🧰 check-dependencies

Use check-dependencies to scan Python files and compare detected imports with the dependencies declared in pyproject.toml.

▶️ Usage

usage: check-dependencies [-h] [--version] [--include-dev] [--verbose] [--provides-from-venv PYTHON_EXECUTABLE] [--missing MODULE,...] [--extra PACKAGE,...] [--provides PACKAGE=MODULE,...] [--include INCLUDE]
                          [--output-format OUTPUT_FORMAT]
                          file_name [file_name ...]

Find undeclared and unused (or all) imports in Python files

positional arguments:
  file_name             Python Source file to analyse

options:
  -h, --help            show this help message and exit
  --version             show program's version number and exit
  --include-dev         Include dev dependencies
  --verbose             Show every import of a package
  --provides-from-venv PYTHON_EXECUTABLE
                        Path to the virtual environment's Python executable
                        (for example, .venv/bin/python) to include all packages
                        installed in it as provides.
  --missing MODULE,...  Comma separated list of requirements known to be missing.
                        
                        Assume they are part of the requirements.
                        Can be specified multiple times.
                        Toml Key: [tool.check-dependencies] known-missing=[]
  --extra PACKAGE,...   Comma separated list of requirements known to not be imported.
                        
                        Assume they are not part of the requirements. This can be plugins or
                        similar that affect the package but are not imported explicitly.
                        Can be specified multiple times.
                        Toml Key: [tool.check-dependencies] known-extra=[]
  --provides PACKAGE=MODULE,...
                        Map a package name to its module (import) name for packages whose import
                        name differs from the package name. Can be specified multiple times.
                        
                        E.g. --provides Pillow=PIL --provides PyJWT=jwt.
                        The package name is normalized (case-insensitive, hyphens and underscores
                        are equivalent), so Pillow=PIL, pillow=PIL and PIL-ow=PIL are all the same.
                        Toml Key: [tool.check-dependencies.provides]
  --include INCLUDE, -I INCLUDE
                        Additional config files to include.
                        Can be specified multiple times. E.g. --include check-dependencies.toml.
                        Toml Key: [tool.check-dependencies] includes=[]
  --output-format OUTPUT_FORMAT
                        The format to use for printing diagnostic messages
                        
                        Possible values:
                        - full:     Print all imports, including correct ones.
                        - concise:  Print only problematic imports (missing or extra)
                        - github:   Print only problematic imports in a format suitable
                            for GitHub Actions annotations

### 📄 Output

The output is a list of imports prefixed with their status.

Default status prefixes:

- `!` - Undeclared import
- `+` - Extra dependency, declared in `pyproject.toml` but not used in the code
- `?` - Dynamic import that could not be resolved.
- `!!` - Could not parse the file (e.g. syntax error)
- `!E` - Could not find associated pyproject.toml file
- ` ` - Correct import (only shown with `--output-format full`)

With `--verbose`, the output includes every matching import together with the
file name and line number where it appears.

Verbose status prefixes:

- `!NA` - Undeclared import
- `+EXTRA` - Extra dependency, declared in `pyproject.toml` but not used in the code
- `?UNKNOWN` - Dynamic import that could not be resolved.
- `!!FILE_ERROR` - Could not parse the file (e.g. syntax error)
- ` OK` - Correct import (only shown with `--output-format full`)

### 📝 Examples

#### Basic usage

▶️ Command:

```shell
check-dependencies project/src/

Example output:

  pandas
! matplotlib
  numpy
# Project project/pyproject.toml
+ requests

Add known extra requirements

Use this when dependencies affect the application but are not imported directly in the codebase, such as plugins.

• ▶️ Command:

  check-dependencies --extra snowflake-sqlalchemy project/src
  

• 📄 pyproject.toml:

  [tool.check-dependencies]
  known-extra = [ "snowflake-sqlalchemy" ]
  

Translate package names

Some packages have different distribution and import names, for example Pillow is imported as PIL.

• ▶️ Command:

  check-dependencies --provides Pillow=PIL --provides PyJWT=jwt project/src
  

• 📄 pyproject.toml:

  [tool.check-dependencies.provides]
  Pillow = "PIL"
  PyJWT = "jwt"
  

Support for namespaces

Supports PEP 420-style namespace-package imports where imports look like company.package_name, with a declared dependency like company.package_name (or the import is covered via a [tool.check-dependencies.provides] mapping).

• 📄 pyproject.toml:

  [project]
  dependencies = [ "company.package_name" ]
  

• 📄 pyproject.toml:

  dependencies = [ "package_name" ]
  [tool.check-dependencies.provides]
  package_name = ["company.package_name"]
  

Add known missing requirements

Use this when imports are expected to be missing from the dependency list, but should not be reported.

• ▶️ Command:

  check-dependencies --missing numpy project/src
  

• 📄 pyproject.toml:

  [tool.check-dependencies]
  known-missing = [ "numpy" ]
  

Map optional dependencies to paths

By default, optional dependencies are checked just like regular dependencies. If you want to check them only for certain files, you can configure them in the pyproject.toml file under [tool.check-dependencies.optional-dependencies] with a mapping of dependency groups to path prefixes.

This is only configurable via pyproject.toml and not via CLI arguments.

• 📄 pyproject.toml:

  [project.optional-dependencies]
  optional_dependency_group = ["optional_dependency_1", "optional_dependency_2"]
  another_optional_dependency_group = ["optional_dependency_3"]
  
  [tool.check-dependencies.optional-dependencies]
   optional_dependency_group = ["src/project/option_1/"]
   another_optional_dependency_group = ["src/option_2.py"]
  

Include additional config file

Use an additional config file to provide extra dependencies, missing dependencies, or provides mappings.

This is especially useful in monorepos where multiple packages share a common configuration file.

• ▶️ Command:

  check-dependencies --include ../global-check-dependencies.toml project/src/
  

• 📄 pyproject.toml:

  [tool.check-dependencies]
  includes = [ "../global-check-dependencies.toml" ]
  

Include dev dependencies

• ▶️ Command:

  check-dependencies --include-dev project/tests/
  

Include provides from virtual environment

Read package-to-import mappings from a virtual environment and include them in the check.

• ▶️ Command:

  check-dependencies --provides-from-venv .venv/bin/python project/src/
  

Output all dependencies

Show all detected dependencies, including the correct ones.

In the following example, pandas is declared and used, requests is declared but unused, and numpy is used but not declared.

check-dependencies --output-format full project/src/

Example output:

  pandas
! numpy
+ requests

Verbose output

Show each import together with its status, file name, and line number.

check-dependencies --verbose project/src/

Example output:

# ALL=False
# INCLUDE_DEV=False
# EXTRA pytest
# EXTRA toml
# EXTRA tomllib
# MISSING check_dependencies
# MISSING toml
# MISSING tomllib
!!FILE_ERROR project/src/broken.py
!NA matplotlib project/src/main.py:4

##### project/pyproject.toml ###
# Dependencies in config file not used in application:
+EXTRA requests

Combine verbose and all

Show all imports, including correct ones, with file names and line numbers.

check-dependencies --verbose --output-format full project/src/

Example output:

# ALL=True
# INCLUDE_DEV=False
# EXTRA pytest
# EXTRA toml
# EXTRA tomllib
# MISSING check_dependencies
# MISSING toml
# MISSING tomllib
 OK project/src/data.py:5 pandas
 OK project/src/main.py:3 pandas
 OK project/src/plotting.py:4 pandas
!NA project/src/plotting.py:5 matplotlib
 OK project/src/plotting.py:6 numpy

### Dependencies in config file not used in application:
# Config file: project/pyproject.toml
+EXTRA requests

⚙️ Configuration

Configuration is read from pyproject.toml.

[tool.check-dependencies]
known-missing = [
    "undeclared_package",
    "another_package"
]
known-extra = [
    "package_as_extra_for_another_package",
    "yet_another_package"
]

[tool.check-dependencies.provides]
# Maps package name (as declared in dependencies) -> import/module name
Pillow = "PIL"
PyJWT = "jwt"
pyshp = "shapefile"
foxtrox = ["fox", "trox"]  # This package provides both `import fox` and `import trox`, but the package name is `foxtrox`
[tool.check-dependencies]
includes = [
  "check-dependencies.toml",
  "../../common-provides.toml"
]

🚦 Exit codes

• 0: No missing or superfluous dependencies found
• 2: Missing dependencies found (used, but not declared in pyproject.toml)
• 4: Extra dependencies found (declared in pyproject.toml, but unused)
• 6: Both missing and superfluous dependencies found
• 8: Could not find associated pyproject.toml file
• 16: Could not parse source file(s)
• 1: Another error occurred

✍️ dependency-writer

Use dependency-writer to generate or update [tool.check-dependencies.provides] mappings from an existing Python environment.

This is useful for generating the initial config file or refreshing it after dependency changes.

Combined with the includes setting in [tool.check-dependencies], it can also be used to generate a shared [tool.check-dependencies.provides] mapping for a monorepo.

If you install the package yourself and want to use dependency-writer, make sure the optional write extra is installed because this command depends on tomlkit.

▶️ Usage

usage: dependency-writer [-h] --python PYTHON --config CONFIG

options:
  -h, --help           show this help message and exit
  --python, -p PYTHON  Python executable to check.
  --config, -c CONFIG  Location of toml config file.

📝 Examples

Write to pyproject.toml

The following command updates the [tool.check-dependencies.provides] table of pyproject.toml with all mappings found in the virtual environment.

• ▶️ Command:

dependency-writer -p .venv/bin/python -c pyproject.toml

Write a global provides file for a monorepo

• ▶️ Command:

dependency-writer -p apps/my-app/.venv/bin/python -c ./check-dependencies.toml

This requires an includes = [...] entry under [tool.check-dependencies] in the application's pyproject.toml so that the generated config file is included:

[tool.check-dependencies]
includes = [ "../../check-dependencies.toml" ]

🛠️ Development

See CONTRIBUTING.md for development setup and guidelines.