LLM Providers

ClearPR supports 5 LLM providers out of the box. Switch providers with a single environment variable.

Provider Configuration

Set LLM_PROVIDER in your .env file:

Provider	LLM_PROVIDER	Default Model	API Key Required
Anthropic Claude	anthropic	claude-sonnet-4-20250514	Yes
OpenAI	openai	gpt-4o	Yes
Ollama	ollama	llama3	No
Mistral	mistral	mistral-large-latest	Yes
Google Gemini	gemini	gemini-2.5-pro	Yes
Local Agent	agent	claude-code	No (bearer token)

Anthropic Claude (Default)

LLM_PROVIDER=anthropic
LLM_API_KEY=sk-ant-api03-...
# LLM_MODEL=claude-sonnet-4-20250514  # optional

Best review quality. ClearPR's prompts are optimized for Claude.

OpenAI

LLM_PROVIDER=openai
LLM_API_KEY=sk-proj-...
# LLM_MODEL=gpt-4o

Also works with Azure OpenAI by setting LLM_BASE_URL:

LLM_PROVIDER=openai
LLM_API_KEY=your-azure-key
LLM_BASE_URL=https://your-resource.openai.azure.com/openai/deployments/your-deployment

Ollama (Local / Self-Hosted)

LLM_PROVIDER=ollama
LLM_MODEL=llama3
LLM_BASE_URL=http://localhost:11434/v1

No API key needed. Runs entirely on your hardware. Best for data-sensitive environments.

TIP

Make sure Ollama is running and the model is pulled:

ollama pull llama3

Mistral

LLM_PROVIDER=mistral
LLM_API_KEY=your-mistral-key
# LLM_MODEL=mistral-large-latest

Google Gemini

LLM_PROVIDER=gemini
LLM_API_KEY=your-google-ai-key
# LLM_MODEL=gemini-2.5-pro

Local Agent (Claude Code)

Route reviews through a local Claude Code agent that exposes POST /trigger and runs claude -p non-interactively. This reuses an existing Claude subscription on the host instead of an API key.

LLM_PROVIDER=agent
LLM_BASE_URL=http://host.docker.internal:8765
LLM_API_KEY=your-agent-bearer-token

LLM_BASE_URL points at the agent's host and port; ClearPR appends /trigger. LLM_API_KEY is sent as Authorization: Bearer <token>. The agent is expected to return { ok, result: { result, usage, modelUsage } }, where result.result is the review text.

TIP

When ClearPR runs in Docker and the agent runs on the host, use host.docker.internal so the container can reach it, and add extra_hosts: ["host.docker.internal:host-gateway"] to the app service in docker-compose.yml.

Custom Model

Override the default model for any provider:

LLM_PROVIDER=openai
LLM_MODEL=gpt-4-turbo

Embeddings (PR memory)

Separate from the LLM, the PR-memory feature embeds past review comments so it can flag repeat issues. Pick the embedding provider with EMBEDDING_PROVIDER:

Provider	EMBEDDING_PROVIDER	Default model	Dimensions	API key
Voyage AI	voyage	voyage-3-lite	512	Yes (VOYAGE_API_KEY)
Local	local	Xenova/all-MiniLM-L6-v2	384	No

Local runs a sentence-transformers model in-process via transformers.js, no API key, fully on-box. It downloads the model once (cache it on a volume with EMBEDDING_CACHE_DIR):

EMBEDDING_PROVIDER=local
EMBEDDING_MODEL=Xenova/all-MiniLM-L6-v2
EMBEDDING_DIMENSIONS=384
EMBEDDING_CACHE_DIR=/app/models

WARNING

EMBEDDING_DIMENSIONS must match the model (512 for voyage-3-lite, 384 for all-MiniLM-L6-v2). Local embeddings require a glibc-based image (the shipped image is node:slim); they will not load on Alpine. If you leave EMBEDDING_PROVIDER unset/voyage with no key, PR memory is silently skipped and the rest of the review still works.

Architecture

All providers extend the same LlmProviderPort abstract class. The LlmProviderRegistry selects the right adapter at startup based on LLM_PROVIDER. Adding a new provider means creating one adapter file - no changes to domain logic.

LlmProviderPort (abstract)
├── AnthropicLlmAdapter
├── OpenAiLlmAdapter
├── OllamaLlmAdapter
├── MistralLlmAdapter
├── GeminiLlmAdapter
└── AgentLlmAdapter