Providers

Providers

Prompt2PR supports four LLM providers. Providers are interchangeable — the action behaves the same regardless of which one you choose. Pick based on your preference, existing subscriptions, or model availability.

---

Mistral

Mistral AI offers high-quality models with competitive pricing.

Default model: mistral-large-latest

Setup

1. Sign up at console.mistral.ai
2. Go to API Keys and create a new key
3. Add MISTRAL_API_KEY as a GitHub Secret in your repository
4. Set provider: mistral in your workflow

Workflow

- uses: davd-gzl/Prompt2PR@v1
  with:
    prompt: 'Your prompt here'
    provider: mistral
  env:
    MISTRAL_API_KEY: $
    GITHUB_TOKEN: $

---

OpenAI

OpenAI provides GPT models with broad language and code understanding.

Default model: gpt-4o

Setup

1. Sign up at platform.openai.com
2. Go to API Keys and create a new secret key
3. Add OPENAI_API_KEY as a GitHub Secret in your repository
4. Set provider: openai in your workflow

Workflow

- uses: davd-gzl/Prompt2PR@v1
  with:
    prompt: 'Your prompt here'
    provider: openai
  env:
    OPENAI_API_KEY: $
    GITHUB_TOKEN: $

---

Anthropic

Anthropic provides Claude models known for detailed, careful analysis.

Default model: claude-sonnet-4-20250514

Setup

1. Sign up at console.anthropic.com
2. Go to API Keys and create a new key
3. Add ANTHROPIC_API_KEY as a GitHub Secret in your repository
4. Set provider: anthropic in your workflow

Workflow

- uses: davd-gzl/Prompt2PR@v1
  with:
    prompt: 'Your prompt here'
    provider: anthropic
  env:
    ANTHROPIC_API_KEY: $
    GITHUB_TOKEN: $

---

GitHub Models

GitHub Models lets you use LLMs directly through GitHub’s infrastructure. No external API key required — it uses the built-in GITHUB_TOKEN.

Default model: openai/gpt-4o

Prerequisites

• A GitHub Copilot subscription
• Access to GitHub Models

Setup

1. No additional secrets needed
2. Add models: read to your workflow permissions
3. Set provider: github in your workflow
4. Use models in publisher/model-name format

Available Models

GitHub Models hosts models from multiple publishers. Use the publisher/model-name format:

• openai/gpt-4o
• anthropic/claude-sonnet-4.5
• And more — check the GitHub Models marketplace

Workflow

permissions:
  contents: write
  pull-requests: write
  models: read

jobs:
  prompt2pr:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: davd-gzl/Prompt2PR@v1
        with:
          prompt: 'Your prompt here'
          provider: github
          model: openai/gpt-4o
        env:
          GITHUB_TOKEN: $

Important: The models: read permission is required. Without it, the action will fail with an authentication error. {: .notice–warning }

---

Switching Providers

Since providers are interchangeable, switching is straightforward. Change two things:

1. The provider: input
2. The API key in the env: block

# Before (Mistral)
provider: mistral
env:
  MISTRAL_API_KEY: $

# After (OpenAI)
provider: openai
env:
  OPENAI_API_KEY: $

The action handles all differences in API format, authentication, and response parsing internally.

Custom Endpoints

Use the base_url input to point any provider at a custom endpoint:

with:
  provider: openai
  base_url: 'https://my-proxy.example.com/v1'

This is useful for API proxies, self-hosted models, or testing with mock servers.

---

Next: Examples