The last README you'll ever write by hand.
AI-powered README generation β for every language, framework, and developer.
β― pip install readmeai && readmeai --api openai -r https://github.com/your/project
β README.md generated in 12.3s
π Quickstart Β· β¨ Features Β· βοΈ Configuration Β· πΌ Gallery Β· π€ Contributing
ReadmeAI scans your entire codebase β local directory or remote URL β and generates a beautifully structured, production-quality README.md in seconds, powered by state-of-the-art language models.
No blank-page paralysis. No boilerplate copy-paste. No documentation debt.
| What you do | What ReadmeAI does |
|---|---|
| Point at a repo | Analyzes all source files, configs, and dependencies |
| Choose an LLM | Generates project intro, features, architecture, setup guides |
| Run one command | Outputs a fully formatted, customized README.md |
π― Designed for simplicity, customization, and developer productivity.
Every element of your README is yours to control β headers, badge colors and styles, logos, emoji packs, and navigation layouts. Ship docs that match your project's personality.
| Style | Preview | CLI Flag |
|---|---|---|
| Classic | ![]() |
--header-style classic |
| Modern | ![]() |
--header-style modern |
| Compact | ![]() |
--header-style compact |
| Console | ![]() |
--header-style console |
| SVG Banner | ![]() |
--header-style banner |
Switch between the world's leading LLMs with a single flag. Or run entirely offline.
| Provider | Flag | Best For | Cost |
|---|---|---|---|
| OpenAI | --api openai |
Industry-leading general quality | Paid |
| Anthropic Claude | --api anthropic |
Advanced reasoning, nuanced docs | Paid |
| Google Gemini | --api gemini |
Multimodal, cutting-edge | Paid |
| Ollama | --api ollama |
100% local, any open model | Free |
| Offline Mode | --api offline |
Structure only, no AI | Free |
ReadmeAI produces complete, professional documentation sections β not templates, not stubs.
βΈ Click to see what gets generated
| Section | Description |
|---|---|
| Project Introduction | Value proposition, purpose, target audience β crafted by AI |
| Feature Table | Structured breakdown of your project's capabilities |
| Architecture Overview | Directory tree with module-level summaries |
| Getting Started | Auto-detected dependencies, install steps, environment setup |
| Usage Examples | Real commands and code snippets pulled from your codebase |
| Testing Guide | Test execution instructions, framework-specific commands |
| Roadmap | Suggested milestones and future enhancements |
| Contributing | Contribution workflow, standards, and community links |
| License & Acknowledgments | Auto-detected license, credits, and ecosystem attribution |
| β Repository Platforms | β Languages & Runtimes | β Frameworks |
|---|---|---|
| GitHub | Python, Go, Java | FastAPI, Django |
| GitLab | Kotlin, Rust, C++ | React, Vue, Svelte |
| Bitbucket | TypeScript, JS | Docker, K8s, Terraform |
| Local filesystem | Ruby, Swift, PHP | Spring, Rails, Laravel |
Control exactly which files ReadmeAI analyzes. Same syntax as .gitignore β add patterns to exclude generated files, vendor directories, or anything you don't want documented.
| Requirement | Minimum | Notes |
|---|---|---|
| Python | 3.9+ | python.org |
| LLM API key | β | OpenAI / Anthropic / Gemini β or use Ollama / offline for free |
Choose your install method:
# pip β recommended for most users
β― pip install -U readmeai
# pipx β keeps readmeai isolated from other projects
β― pipx install readmeai
# uv β fastest install available
β― uv tool install readmeai
# Docker β zero-dependency, runs anywhere
β― docker pull zeroxeli/readme-ai:latestβΈ Build from source
β― git clone https://github.com/eli64s/readme-ai
β― cd readme-ai
β― pip install -r setup/requirements.txt
# Or with Poetry:
β― poetry install
# Or with the setup script:
β― bash setup/setup.shNote β Anthropic or Gemini users: install the provider extras:
pip install "readmeai[anthropic]" pip install "readmeai[google-generativeai]" pip install "readmeai[anthropic,google-generativeai]"
# OpenAI
β― export OPENAI_API_KEY=<your_key>
# Anthropic
β― export ANTHROPIC_API_KEY=<your_key>
# Google Gemini
β― export GOOGLE_API_KEY=<your_key>
# Ollama β no key needed, just start the server
β― export OLLAMA_HOST=127.0.0.1 && ollama serveMinimal β start here:
β― readmeai --api openai -r https://github.com/your/projectSwitch providers in seconds:
# Anthropic Claude
β― readmeai --api anthropic -m claude-3-5-sonnet-20240620 -r https://github.com/your/project
# Google Gemini
β― readmeai --api gemini -m gemini-1.5-flash -r https://github.com/your/project
# Local Ollama (free, private)
β― readmeai --api ollama --model llama3.2 -r https://github.com/your/project
# Fully offline β no AI, no internet
β― readmeai --api offline -r https://github.com/your/projectFull customization β power user mode:
β― readmeai \
--repository https://github.com/your/project \
--api openai \
--model gpt-4 \
--output my-readme.md \
--badge-color A931EC \
--badge-style flat-square \
--header-style compact \
--navigation-style fold \
--logo LLM \
--emojis solar \
--temperature 0.9 \
--tree-depth 2Docker:
β― docker run -it --rm \
-e OPENAI_API_KEY=$OPENAI_API_KEY \
-v "$(pwd)":/app zeroxeli/readme-ai:latest \
--repository https://github.com/your/project \
--api openai
β οΈ The Streamlit app may lag behind the CLI on the latest features. Use the CLI for cutting-edge functionality.
# Install dev + test dependencies
β― uv pip install --dev --group test --all-extras
# Run the unit test suite
β― make test
# Test across Python 3.9 / 3.10 / 3.11 / 3.12 via nox
β― make test-noxAll options are available as CLI flags. Every single one has a sensible default β customize only what you need.
| Flag | Description | Default |
|---|---|---|
--api |
LLM provider (openai Β· anthropic Β· gemini Β· ollama Β· offline) |
offline |
--model / -m |
LLM model name | gpt-3.5-turbo |
--repository / -r |
Repository URL or local path | required |
--output / -o |
Output filename | readme-ai.md |
--header-style |
Layout (classic Β· modern Β· compact Β· console Β· banner) |
classic |
--badge-style |
Badge look (flat Β· flat-square Β· for-the-badge Β· plastic) |
flat |
--badge-color |
Badge color (name or hex, e.g. A931EC) |
0080ff |
--navigation-style |
TOC format (bullet Β· number Β· roman Β· accordion Β· fold) |
bullet |
--logo |
Logo preset or custom for your own image |
blue |
--emojis |
Emoji theme pack (e.g. solar, water, rainbow) |
None |
--temperature |
LLM creativity, 0.0β2.0 | 0.1 |
--tree-max-depth |
Directory tree depth | 2 |
--align |
Header text alignment | center |
β― readmeai --help # full referenceReadmeAI-generated READMEs across a wide range of real-world projects:
| Stack | Repository | Generated README | Highlight |
|---|---|---|---|
| Python | readme-ai | View β | Core project |
| Apache Flink | pyflink-poc | View β | Modern style |
| Streamlit | readme-ai-streamlit | View β | SVG banner |
| Go + Docker | docker-gs-ping | View β | Multi-language |
| FastAPI + Redis | async-ml-inference | View β | ML stack |
| Java | minimal-todo | View β | Compact style |
| Kotlin (Android) | file.io-Android-Client | View β | Mobile app |
| PostgreSQL + DuckDB | buenavista | View β | Data stack |
| Vercel + NPM | github-readme-quotes | View β | Deployment |
| Offline mode | litellm | View β | No API needed |
π¬ Generated something great? Share it in Show & Tell β
-
readmeai v1.0.0β stable, production-ready release with robust docs creation - Extended templates for more project types and programming languages
- VS Code Extension β generate READMEs directly inside your editor
- GitHub Actions integration β automatically update docs on every push
- Badge packs β code coverage, CI status, version, and more
ReadmeAI is open source and community-driven. All contributions are welcome.
- π Contributing Guide β workflow, coding standards, PR process
- π Report a Bug β something broken? let us know
- π‘ Start a Discussion β ideas, questions, feedback
Built on the shoulders of excellent open-source projects:
Copyright Β© 2023β2025 ReadmeAI. Released under the MIT License.








