RepoAgent

Report Abuse

Basic Information

RepoAgent is an open-source, LLM-powered framework designed to automate repository-level code documentation generation for software projects. It analyzes a target git repository, inspects Python source files via AST parsing, and generates structured documentation for individual objects and the overall project hierarchy. The tool detects repository changes such as file additions, deletions, and modifications, maintains a global project structure record, and writes Markdown documentation into a designated folder. It is distributed as a pip package and can be run from the command line, integrated into GitHub Actions, or installed as a pre-commit hook to generate or update docs automatically on commits. It also provides an experimental chat-with-repo extension for repository Q&A and supports local model backends in addition to hosted APIs.

Links

Categorization

App Details

Features
RepoAgent offers automatic change detection in git repositories to track added, removed, or modified files. It performs AST-based code analysis to document functions, classes and other objects, and identifies inter-object invocation relationships to enrich global documentation. The system can replace Markdown content based on code changes and maintain a global JSON project record and Markdown_Docs output folder. It supports multithreaded concurrent generation for performance, a CLI with options to set model, temperature, timeouts, target paths and logging, and integration with GitHub Actions and pre-commit hooks. Additional features include a chat-with-repo prototype for Q&A, configurable prompts and templates, and support for local models and different model endpoints.
Use Cases
RepoAgent reduces manual effort required to create and maintain repository documentation by automating generation and updates, helping teams keep docs consistent with code changes. Pre-commit and CI integration enable documentation to be produced as part of normal development workflows, which eases onboarding and code review. AST-level analysis and inter-object relationship detection produce more precise module and object-level explanations, and the produced Markdown structure and global JSON record facilitate publishing and integration with documentation sites. The chat-with-repo prototype enables interactive code explanation and issue Q&A. The tool has been applied to large codebases and can be tuned via prompts, model selection and CLI configuration to balance cost and quality.

Please fill the required fields*