Troubleshooting CodeMind
A local Python RAG assistant for asking questions about local files and web pages.
Troubleshooting CodeMind
Resolve common environment, model selection, ingestion, and retrieval problems with these checks.
Module main not found: activate the environment and run python -m main with the current directory set to the cloned repository root.
Ollama connection fails or no models appear: start the Ollama service, verify it is reachable at the configured localhost URL, and pull a chat model and the configured embedding model before retrying.
llama.cpp cannot find a model: place a supported .gguf file in the models directory or follow the download prompt. Larger models need more memory and can run slowly on CPU-only hardware.
Answers have no relevant sources: verify that ingestion completed, check /sources, and try a specific question that appears in the source. Reduce chunk size or adjust TOP_K only after testing the current baseline.
URL ingestion fails: confirm the machine has network access and the target page is publicly reachable. Local inference does not make URL fetching offline.