You control which LLM backs the code review session with theDocumentation Index
Fetch the complete documentation index at: https://mintlify.com/goulinkh/code-review-harness/llms.txt
Use this file to discover all available pages before exploring further.
--model flag. CRH uses a provider:model-id format so you can switch between Anthropic, OpenAI, Kimi, OpenRouter, or any OpenAI-compatible endpoint without changing anything else in your workflow.
Format
provider segment determines which API client and key environment variable CRH uses. The model-id segment is passed directly to that provider’s API.
Default: anthropic:claude-sonnet-4-6
API key environment variables
CRH reads the API key from an environment variable whose name matches the provider prefix in upper-case, followed by_API_KEY. For example:
| Provider prefix | Expected env var |
|---|---|
anthropic | ANTHROPIC_API_KEY |
openai | OPENAI_API_KEY |
kimi-coding | KIMI_API_KEY |
openrouter | OPENROUTER_API_KEY |
Set the appropriate environment variable before running
crh review. CRH does not prompt for credentials interactively.Examples
OpenAI-compatible server with a custom base URL
Use--model-base-url to point any OpenAI-compatible provider or local inference server at a custom endpoint. The --model value must be an OpenAI-compatible model when you supply --model-base-url.
Kimi (Moonshot)
SetKIMI_API_KEY and use the kimi-coding provider prefix:
OpenRouter
OpenRouter lets you access models from many providers through a single API key. Supply your OpenRouter key viaOPENROUTER_API_KEY and pass the full model path as the model-id:
Using --model-base-url
--model-base-url overrides the API base URL for the chosen model. It only works with OpenAI-compatible models — that is, any openai:* model or a registered provider whose API type is openai-completions or openai-responses.
Related pages
crh review command
Full syntax and usage examples for the review command.
Flags reference
Complete reference for every flag accepted by
crh review.