lampe-sdk 🔦

Lampe is the SDK part of the montagne-dev (🏔️🏕️🔦) project.

Using lampe-sdk

You can use lampe-sdk as a dependency in your own Python project. To do so:

1. Install the SDK:

   # Using a branch name
   uv add https://github.com/montagne-dev/lampe.git@branch-name --upgrade
   # Using a tag/version
   uv add https://github.com/montagne-dev/lampe.git@v0.1.0
   # Using pypi
   uv add lampe-sdk
   # Using pip
   pip install lampe-sdk

2. Set Up Environment Variables

You must set the following two environment variables before running any CLI commands:

Variable	Default	Purpose
OPENAI_API_KEY	unset	API key for OpenAI LLM provider.
ANTHROPIC_API_KEY	unset	API key for Anthropic LLM provider.

Example .env:

OPENAI_API_KEY="sk-.."
ANTHROPIC_API_KEY="sk-.."

3. Import and Use the SDK

    # Initialize the SDK (required)
    from lampe.describe import PRDescriptionWorkflow, PRDescriptionStartEvent, generate_pr_description
    initialize()
   
    # Import and use SDK features
    from lampe.describe import PR
    workflow = PRDescriptionWorkflow(truncation_tokens=truncation_tokens, timeout=timeout, verbose=verbose)
    result = await workflow.run(
        start_event=PRDescriptionStartEvent(
            pr_title=pull_request.title,
            repository=repository,
            pull_request=pull_request,
            files_exclude_patterns=files_exclude_patterns,
        )
    )
    # Or using entrypoint
    data = {...} # Add required informations of PRDescriptionInput
    input = PRDescriptionInput.model_validate(data)
   
    result = asyncio.run(
        generate_pr_description(
            repository=input.repository,
            pull_request=input.pull_request,
            files_exclude_patterns=input.files_exclude_patterns,
            files_reinclude_patterns=input.files_reinclude_patterns,
        )
    )
    print(result.description)

The SDK must be initialized before use. The initialization will set up logging and telemetry (if configured).

Documentation

uv run mkdocs serve
Serving on http://127.0.0.1:8000/
# build the documentation
uv run mkdocs build

Open a bowser to http://127.0.0.1:8000/ to view the docs

LLM Configuration

Configure the LITELLM proxy environment variables in your .env file to enable LLM calls. Multiple models are in used in this project from different providers. You must have a valid key from all those providers to be able to run the lampe-sdk review:

• Anthropic
• OpenAI

1. Get API Credentials: Go to each platform create an account or signin, then get an API secret key.

2, Store it in .env: You will then need to create the following environment variable, so litellm can reach the provider for you.

• ANTHROPIC_API_KEY
• OPENAI_API_KEY

GitHub Action

The Lampe SDK includes a GitHub Action that makes it easy to integrate PR description generation and code review into your GitHub workflows.

Quick Start

Add this to your .github/workflows/pr-description.yml:

name: Generate PR Description
on:
  pull_request:

jobs:
  describe:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
        with:
          ref: ${{ github.head_ref }}
      - uses: montagne-dev/lampe@main
        with:
          command: describe
        env:
          OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}

Features

• Automatic Git History Fetching: Includes merge-base functionality to ensure all PR commits are available
• Multiple Commands: Support for describe, review, and healthcheck commands
• Flexible Configuration: Extensive options for customization
• Multiple Output Providers: Support for console, GitHub, GitLab, and Bitbucket outputs
• Two Generation Variants: Default and agentic modes

Advanced Usage

name: Lampe Analysis
on:
  pull_request:

jobs:
  analyze:
    runs-on: ubuntu-latest
    env:
      OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
      LAMPE_GITHUB_APP_ID: ${{ secrets.LAMPE_GITHUB_APP_ID }}
      LAMPE_GITHUB_APP_PRIVATE_KEY: ${{ secrets.LAMPE_GITHUB_APP_PRIVATE_KEY }}
    steps:
      - uses: actions/checkout@v4
        with:
          ref: ${{ github.head_ref }}

      # Health check
      - name: Health Check
        uses: montagne-dev/lampe@main
        with:
          command: healthcheck

      # Generate description
      - name: Generate Description
        uses: montagne-dev/lampe@main
        with:
          command: describe
          variant: agentic
          output: github

Required Secrets

Set these in your repository settings:

• OPENAI_API_KEY: Your OpenAI API key
• ANTHROPIC_API_KEY: Your Anthropic API key (optional, for agentic mode)
• LAMPE_GITHUB_APP_ID: GitHub App ID (optional, for GitHub integration)
• LAMPE_GITHUB_APP_PRIVATE_KEY: GitHub App private key (optional, for GitHub integration)

For detailed documentation, see Github action page in the documentation.

Bitbucket Pipeline

https://support.atlassian.com/bitbucket-cloud/docs/pipeline-start-conditions/#Pull-Requests