Section 4: Usage & Features¶

Welcome to Section 4 of the OSA FAQ! This section covers everything you need to know about running OSA, understanding its features, and maximizing its capabilities for your repository improvement workflow.

4.1 How do I run OSA?¶

Running OSA is straightforward once you have it installed and configured. Here's the complete workflow:

Basic Run (Minimum Required):

# Set your environment variables first
export GIT_TOKEN="ghp_..."
export OPENAI_API_KEY="sk-..."  # If using OpenAI

# Run OSA with repository URL
python -m osa_tool.run -r https://github.com/username/repository

Running with Docker:

docker run --env-file .env osa_tool:latest \
  -r https://github.com/username/repository

4.2 What command-line arguments are available?¶

OSA provides extensive CLI arguments for customization. Here's the complete reference:

Required Arguments:

Flag	Description	Example	Mandatory
`-r, --repository`	URL of GitHub/GitLab/Gitverse repository	`-r https://github.com/aimclub/OSA`	✅ Yes

Core Configuration:

Flag	Description	Default	Example
`-b, --branch`	Target branch name	Default branch	`-b develop`
`-o, --output`	Output directory path	Current directory	`-o ./results`
`-m, --mode`	Operation mode	`auto`	`--mode basic`
`--config-file`	Path to TOML configuration	None	`--config-file config.toml`

LLM Configuration:

Flag	Description	Default	Example
`--api`	LLM provider	`openai`	`--api ollama`
`--base-url`	API endpoint URL	`https://openrouter.ai/api/v1`	`--base-url http://localhost:11434`
`--model`	LLM model name	`gpt-3.5-turbo`	`--model llama3.2:3b`
`--temperature`	Sampling temperature (0-1)	`0.05`	`--temperature 0.3`
`--top_p`	Nucleus sampling probability	`0.95`	`--top_p 0.9`
`--max_tokens`	Max output tokens	`4096`	`--max_tokens 2048`
`--context_window`	Total context window	`16385`	`--context_window 8192`

Task-Specific Models (when --use-single-model=false):

Flag	Purpose	Example
`--model-docstring`	Model for docstring generation	`--model-docstring codellama:13b`
`--model-readme`	Model for README generation	`--model-readme gpt-4o`
`--model-validation`	Model for code validation	`--model-validation llama3.1:8b`
`--model-general`	Model for general tasks	`--model-general gemma3:27b`

Repository Interaction:

Flag	Description	Default	Example
`--no-fork`	Skip creating fork	`False`	`--no-fork`
`--no-pull-request`	Skip creating PR	`False`	`--no-pull-request`
`--delete-dir`	Delete cloned repo after processing	`disabled`	`--delete-dir`

Special Features:

Flag	Description	Example
`--attachment`	Path/URL to PDF/.docx paper	`--attachment ./paper.pdf`
`--convert-notebooks`	Convert Jupyter notebooks	`--convert-notebooks ./notebooks/`
`--generate-workflows`	Generate CI/CD pipelines	`--generate-workflows`

Help:

# View all available arguments
python -m osa_tool.run --help

4.3 How does README generation work?¶

OSA's README generation is one of its core features, creating comprehensive documentation automatically.

README Sections Generated:

Section	Content	Auto-Generated
Project Title	Repository name + description	✅
Badges	Build status, license, version, etc.	✅
Overview	Project purpose and goals	✅
Core Features	Key functionality list	✅
Installation	Step-by-step setup instructions	✅
Quick Start	Basic usage examples	✅
Documentation	Links to docs, API reference	✅
Examples	Code snippets and demos	✅
Contributing	How to contribute guidelines	✅
License	License information	✅
Acknowledgments	Credits and citations	✅

Standard vs Article-Style README:

Feature	Standard README	Article-Style README
Trigger	Default	`--attachment <paper>`
Overview	Project-focused	Research paper-focused
Content Section	Features list	Paper methodology
Algorithms	Optional	Detailed explanation
Citations	Optional	Required
Best For	General projects	Research publications

4.4 How does documentation generation work?¶

OSA generates comprehensive code documentation through a sophisticated multi-stage pipeline.

What Gets Documented:

Component	Documentation Generated
Modules	File-level descriptions, purpose, exports
Classes	Class purpose, attributes, methods overview
Methods	Parameters, return values, behavior, exceptions
Functions	Purpose, arguments, returns, examples
Imports	Dependency relationships mapped

Documentation Output Locations:

project/
├── docs/
│   ├── index.md           # Main documentation page
│   ├── api/               # API reference
│   │   ├── module1.md
│   │   └── module2.md
│   └── examples.md        # Usage examples
├── mkdocs.yml             # MkDocs configuration
└── .github/workflows/
    └── docs-deploy.yml    # Auto-deployment workflow

4.5 How does CI/CD workflow generation work?¶

OSA automatically creates CI/CD pipelines tailored to your repository platform and needs.

Generated Workflow Components:

Component	Purpose	Tools Used
Unit Tests	Run automated tests	pytest, unittest
Code Formatting	Ensure consistent style	Black, autopep8
Linting	Check code quality	flake8, PEP8
Type Checking	Validate type hints	mypy, pyright
Documentation	Build and deploy docs	MkDocs, Sphinx
Package Publish	Upload to PyPI	twine, poetry
Code Coverage	Track test coverage	coverage.py, Codecov

Workflow Customization Options:

Flag	Description	Example
`--use-poetry`	Enable Poetry dependency management	`--use-poetry`
`--branches`	Specify trigger branches	`--branches main,develop`
`--codecov-token`	Codecov integration token	`--codecov-token <token>`
`--include-codecov`	Enable coverage reporting	`--include-codecov`

Repository Structure After CI/CD Setup:

repository/
├── .github/
│   └── workflows/
│       ├── ci.yml           # Main CI pipeline
│       ├── docs-deploy.yml  # Documentation deployment
│       └── publish.yml      # PyPI publication
├── tests/                   # Standardized test directory
│   ├── __init__.py
│   └── test_main.py
├── examples/                # Standardized examples directory
│   ├── basic_usage.py
│   └── advanced_demo.py
└── .gitlab-ci.yml           # GitLab alternative

4.6 Can OSA work with research paper-based projects?¶

Yes! This is one of OSA's unique strengths. It was originally designed specifically for research repositories.

Research Project Support:

Feature	Description	Benefit
Article-Style README	README generated from paper content	Connects code to research
PDF Attachment	Upload paper PDF for context	Accurate methodology documentation
Algorithm Section	Explains implemented methods	Reproduces research claims
Citation Integration	Auto-generates citation section	Proper academic attribution
Reproducibility Focus	Emphasizes setup and usage	Enables result verification

Why Research Projects Need OSA:

Common Problem	OSA Solution
❌ Code shared without README	✅ Auto-generated article-style README
❌ No connection to paper	✅ Paper content integrated into docs
❌ Missing implementation details	✅ Algorithm section from paper
❌ No reproducibility instructions	✅ Clear setup and usage guide
❌ Missing license	✅ Appropriate license added
❌ No CI/CD for validation	✅ Automated testing workflows

Research Repository Example:

Before OSA:

research-code/
├── main.py          # No docstrings
├── utils.py         # No documentation
└── results/         # No README

After OSA:

research-code/
├── README.md        # Article-style with paper summary
├── main.py          # Full docstrings
├── utils.py         # Full docstrings
├── docs/            # MkDocs documentation
├── tests/           # Automated tests
├── examples/        # Usage examples
├── LICENSE          # License file
├── CITATION.cff     # Citation file
└── .github/workflows/
    └── ci.yml       # CI/CD pipeline

Paper Integration Workflow:

1. Upload paper PDF → 2. OSA extracts key information → 
3. Generates article-style README → 4. Links code to methodology →
5. Creates reproducibility guide → 6. Adds citation information

4.7 How do I use the `--attachment` option for papers?¶

The --attachment option enables article-style README generation based on research papers.

Supported Formats:

Format	Local Path	URL	Notes
PDF	`./paper.pdf`	`https://.../paper.pdf`	✅ Fully supported
DOCX	`./paper.docx`	—	⚠️ Limited support
LaTeX	`./paper.tex`	—	⚠️ Limited support

Using Local Files:

# Standard local file
python -m osa_tool.run \
  -r https://github.com/username/repo \
  --attachment ./papers/research_paper.pdf

# With absolute path
python -m osa_tool.run \
  -r https://github.com/username/repo \
  --attachment /home/user/papers/paper.pdf

Using URLs:

# Direct PDF URL
python -m osa_tool.run \
  -r https://github.com/username/repo \
  --attachment https://arxiv.org/pdf/2401.12345.pdf

# Research gate, ACM, IEEE links (if publicly accessible)
python -m osa_tool.run \
  -r https://github.com/username/repo \
  --attachment https://dl.acm.org/doi/pdf/10.1145/xxxxxxx

Using with Docker:

# Method 1: Copy file into image before building
cp paper.pdf ./docker/papers/
docker build -t osa_tool .
docker run --env-file .env osa_tool:latest \
  -r https://github.com/username/repo \
  --attachment /app/OSA/papers/paper.pdf

# Method 2: Volume mounting (recommended)
docker run --env-file .env \
  -v $(pwd)/papers:/app/OSA/papers \
  osa_tool:latest \
  -r https://github.com/username/repo \
  --attachment /app/OSA/papers/paper.pdf

4.8 What are the different operation modes (basic/auto/advanced)?¶

OSA offers three modes to fit different needs:

Mode	Description	Best For
🟢 Basic	Predefined improvements: README, report, community files, tests/examples folders	Quick standard improvements
🔵 Auto (Default)	LLM analyzes repo and creates customized improvement plan for approval	Most users, balanced approach
🟠 Advanced	Full manual control over every step and configuration	Power users, specific requirements

python -m osa_tool.run \
  -r https://github.com/username/repo \
  --mode advanced

Experimental Features:

Feature	Status	Description
Conversational Mode	🚧 Under Development	Natural language requests via CLI
Multi-Agent System	🚧 Under Development	Specialized agents for different tasks

4.9 Does OSA create pull requests automatically?¶

Yes! By default, OSA automatically creates pull requests with all proposed changes.

4.10 Can I prevent OSA from creating forks/PRs?¶

Yes! You can run OSA in read-only mode without creating forks or pull requests.

Options to Disable Automatic Changes:

Flag	Effect	Use Case
`--no-fork`	Skip fork creation	Read-only analysis of public repos
`--no-pull-request`	Skip PR creation	Review changes locally first
Both flags	Local changes only	Full control over what gets pushed

Read-Only Mode:

# Analyze without any remote repository modifications
python -m osa_tool.run \
  -r https://github.com/username/repo \
  --no-fork \
  --no-pull-request

4.11 Which repository platforms are supported?¶

OSA supports multiple repository hosting platforms.

Supported Platforms:

Platform	Support Level	Features	Token Type
GitHub	✅ Full	All features	`GITHUB_TOKEN` / `GIT_TOKEN`
GitLab.com	✅ Full	All features	`GITLAB_TOKEN` / `GIT_TOKEN`
GitLab Self-Hosted	✅ Full	All features	`GITLAB_TOKEN` / `GIT_TOKEN`
Gitverse	✅ Full	All features	`GITVERSE_TOKEN` / `GIT_TOKEN`

CI/CD Platform Differences:

Feature	GitHub	GitLab	Gitverse
Workflow Location	`.github/workflows/`	`.gitlab-ci.yml`	Platform-specific
Workflow Format	YAML (Actions)	YAML (CI/CD)	YAML
PR Terminology	Pull Request	Merge Request	Pull Request
Branch Protection	Supported	Supported	Supported
Pages Deployment	GitHub Pages	GitLab Pages	Platform Pages

Platform Detection:

OSA automatically detects the platform from the repository URL:

URL Pattern	Detected Platform
`github.com/*`	GitHub
`gitlab.com/*`	GitLab
`gitlab./`	GitLab (self-hosted)
`gitverse.io/*`	Gitverse

Token Permissions by Platform:

Platform	Required Scopes
GitHub	`repo`, `workflow`
GitLab	`api`, `write_repository`
Gitverse	`repo`, `write`

4.12 Can I process Jupyter notebooks?¶

Yes! OSA can convert Jupyter notebooks to Python scripts for processing.

Using Notebook Conversion:

# Convert specific notebooks
python -m osa_tool.run \
  -r https://github.com/username/repo \
  --convert-notebooks ./notebooks/tutorial.ipynb

# Convert all notebooks in directory
python -m osa_tool.run \
  -r https://github.com/username/repo \
  --convert-notebooks ./notebooks/

# Multiple notebook paths
python -m osa_tool.run \
  -r https://github.com/username/repo \
  --convert-notebooks ./notebooks/*.ipynb