Skip to content

providers.md

Latest commit

 

History

History
200 lines (136 loc) · 4.17 KB

providers.md

File metadata and controls

200 lines (136 loc) · 4.17 KB

Providers

git-msg supports four LLM providers. All use stdlib net/http — no vendor SDKs are bundled.

Anthropic (Claude)

API reference: docs.anthropic.com

Setup

  1. Create an account at console.anthropic.com
  2. Generate an API key under API Keys
  3. Run the setup wizard:
rm ~/.config/mdstn/git-msg/config.toml
git-msg generate --dry-run
# Select: Anthropic (Claude)
# Enter your key when prompted

Or configure manually:

git-msg config set provider.name anthropic
git-msg config set provider.model claude-haiku-4-5
# Key is stored interactively via wizard, or:
export GIT_MSG_ANTHROPIC_API_KEY=sk-ant-...

Recommended models

Model Speed Quality Notes
claude-haiku-4-5 Fast Good Default — best cost/speed ratio
claude-sonnet-4-5 Medium Great Better for complex diffs
claude-opus-4-5 Slow Best Large context, detailed commits

API endpoint

https://api.anthropic.com/v1/messages

OpenAI (GPT)

API reference: platform.openai.com/docs

Setup

  1. Create an account at platform.openai.com
  2. Generate an API key under API Keys
  3. Configure:
git-msg config set provider.name openai
git-msg config set provider.model gpt-4o-mini
export GIT_MSG_OPENAI_API_KEY=sk-...

Recommended models

Model Speed Quality Notes
gpt-4o-mini Fast Good Default — cheap and capable
gpt-4o Medium Great Better reasoning
o1-mini Slow Best Highest quality, higher cost

API endpoint

https://api.openai.com/v1/chat/completions

Google Gemini

API reference: ai.google.dev

Setup

  1. Go to aistudio.google.com and create an API key
  2. Configure:
git-msg config set provider.name gemini
git-msg config set provider.model gemini-1.5-flash
export GIT_MSG_GEMINI_API_KEY=AIza...

Recommended models

Model Speed Quality Notes
gemini-1.5-flash Fast Good Default — generous free tier
gemini-1.5-pro Medium Great Larger context window
gemini-2.0-flash Fast Great Latest Flash generation

API endpoint

https://generativelanguage.googleapis.com/v1beta/models/<model>:generateContent

The API key is passed as a query parameter (?key=...), not a header.

Ollama (local)

Documentation: ollama.ai

Ollama runs models locally on your machine — no API key required, no data leaves your network.

Setup

  1. Install Ollama from ollama.ai/download
  2. Pull a model:
ollama pull llama3
ollama pull mistral
ollama pull codellama
  1. Configure git-msg:
git-msg config set provider.name ollama
git-msg config set provider.model llama3

Or run the setup wizard — it will query ollama list and show a picker of your installed models.

Listing available models

ollama list

git-msg queries this automatically during the setup wizard.

Custom host

If Ollama is running on a different machine or port:

git-msg config set ollama.host http://192.168.1.50:11434

Recommended models for commit messages

Model Size Notes
llama3 4.7GB Good all-rounder
mistral 4.1GB Fast, concise output
codellama 3.8GB Code-focused
phi4 9.1GB High quality, larger

API endpoint

http://localhost:11434/api/chat

No Authorization header is sent.

Switching providers

For a single run, use --provider:

git-msg generate --provider ollama
git-msg generate --provider anthropic

To change the default permanently:

git-msg config set provider.name gemini
git-msg config set provider.model gemini-1.5-flash

Timeouts

All providers use a 30-second HTTP timeout. If the LLM is slow to respond (large diff, slow network), the request will fail with a timeout error. Reduce diff size by staging fewer files, or use a faster model.