grepai-embeddings-ollama

You can generate local embeddings by configuring Ollama as the embedding provider for GrepAI. This setup enables on-device vector generation and offline indexing, keeping sensitive codebase data entirely private within your environment.

Yes, offline indexing supports both local Ollama instances and custom remote endpoints. You can point GrepAI to your desired server by specifying the endpoint URL in the .grepai/config.yaml configuration file.

To configure GrepAI, set the embedder provider, model, and endpoint in your .grepai/config.yaml file. After updating the configuration, verify that the Ollama server is running before starting the indexing watch process.

The primary prerequisite is a running Ollama server and a compatible embedding model. You must also ensure your .grepai/config.yaml file correctly specifies the provider, model, and endpoint to generate vector embeddings locally.

Yes, on-device semantic search with Ollama is designed for large codebases. By running a watch to index your codebase locally, it handles extensive vector generation while maintaining strict data privacy.

Local embedding configuration usually fails if the Ollama server is not running or if the .grepai/config.yaml file lacks the correct provider, model, and endpoint settings. Verify these configuration requirements to resolve the issue.