Providers

The reviewer supports four LLM providers. Configure your preferred one in .ai-review.yml:

provider: openai   # openai | anthropic | gemini | ollama
model: gpt-4o-mini # provider-specific model name

Provider comparison

Provider	GitHub Secret	Default model	Console
OpenAI	`OPENAI_API_KEY`	`gpt-4o-mini`	platform.openai.com (opens in a new tab)
Anthropic	`ANTHROPIC_API_KEY`	`claude-sonnet-4-20250514`	console.anthropic.com (opens in a new tab)
Gemini	`GEMINI_API_KEY`	`gemini-1.5-flash`	aistudio.google.com (opens in a new tab)
Ollama	(no key — self-hosted)	set via `model:`	ollama.com (opens in a new tab)

OpenAI

Default provider. Uses response_format: json_schema for structured output — the model is constrained to valid JSON matching the review schema with no post-processing needed.

# .ai-review.yml
provider: openai
model: gpt-4o-mini   # or gpt-4o for higher quality

# GitHub Actions workflow
- run: npx -y @giolabsuy/ai-code-reviewer@latest review-pr
  env:
    OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
    GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

Anthropic

Uses explicit JSON formatting instructions appended to the system prompt.

# .ai-review.yml
provider: anthropic
model: claude-sonnet-4-20250514

# GitHub Actions workflow
- run: npx -y @giolabsuy/ai-code-reviewer@latest review-pr
  env:
    ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
    GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

Gemini

Uses explicit JSON formatting instructions appended to the system prompt.

# .ai-review.yml
provider: gemini
model: gemini-1.5-flash   # or gemini-1.5-pro for higher quality

# GitHub Actions workflow
- run: npx -y @giolabsuy/ai-code-reviewer@latest review-pr
  env:
    GEMINI_API_KEY: ${{ secrets.GEMINI_API_KEY }}
    GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

Ollama (self-hosted)

Ollama runs locally and requires no API key, but the service must be running and reachable from the environment where the reviewer runs.

# .ai-review.yml
provider: ollama
model: codellama          # or any model you have pulled
ollamaUrl: http://localhost:11434   # default if omitted

For GitHub Actions, Ollama must be running as a service in the job or reachable via a network URL. For local use, just start ollama serve before running the reviewer.

For most teams, OpenAI gpt-4o-mini is the best starting point — good quality, low cost, and structured output guarantees.

Local Usage Configuration