tldr

i ain't reading all that. im happy for you tho, or sorry that happened

A tool for summarizing GitHub PRs using LLMs.

Features

• Summarize GitHub PRs with OpenAI LLMs
• Chat with an AI about the PR details
• Export PR summaries in Markdown for easy copying to GitHub

Installation

Install with pip

pip install git+https://github.com/wandb/tldr.git

Usage

# Summarize the current PR in the current directory
tldr

# Summarize a PR in a specific repository
tldr /path/to/repo

# Use a specific model (default is gpt-4o)
tldr --model gpt-3.5-turbo

# Start an interactive chat session after summarizing
tldr --chat

# Output summary as Markdown
tldr --md

Run with uvx

To run tldr from the main branch with uvx as a standalone script in an isolated environment, you can run:

uvx --from git+https://github.com/wandb/tldr@main tldr

Requirements

• Python 3.8+
• OpenAI API key (set as OPENAI_API_KEY environment variable)
• GitHub CLI (gh) for accessing PR information
• Git

Project Structure

The project is organized into modules:

• utils.py: Utility functions for file operations and Git
• git.py: Functions for working with Git repositories and PRs
• agent.py: LLM agent functionality for analyzing PRs
• chat.py: Interactive chat functionality
• cli.py: Command-line interface
• __main__.py: Entry point for the package

PR Summarization Agent Loop

This diagram outlines the workflow used by the agent for summarizing pull requests:

flowchart TD
    Start([Start]) --> FetchPR[Fetch PR Data]
    FetchPR --> PRExists{PR Found?}
    PRExists -- No --> ErrorA[Return Error]
    PRExists -- Yes --> ProcessFiles[Process PR Files]
    ProcessFiles --> FilesExist{Files Found?}
    FilesExist -- No --> BasicSummary[Generate Basic Metadata Summary]
    FilesExist -- Yes --> AnalyzeChanges[Analyze File Changes]
    AnalyzeChanges --> ExtractPatterns[Extract Dependency Patterns]
    ExtractPatterns --> SearchDependencies[Search Codebase for Dependencies]
    SearchDependencies --> InitContext[Initialize Context]
    InitContext --> EvalContext[Evaluate Context Sufficiency]
    EvalContext --> Sufficient{Context Sufficient?}
    Sufficient -- No --> MaxIter{Max Iterations?}
    MaxIter -- No --> RunCmd[Run Additional Command]
    RunCmd --> AddOutput[Add Output to Context]
    AddOutput --> EvalContext
    MaxIter -- Yes --> ForceSummary[Force Summary Generation]
    Sufficient -- Yes --> GenSummary[Generate Final Summary]
    ForceSummary --> ReturnSummaryA[Return Summary]
    GenSummary --> ReturnSummaryB[Return Summary]
    BasicSummary --> End([End])
    ErrorA --> End
    ReturnSummaryA --> End
    ReturnSummaryB --> End

Loading