Description

openclaw doctor reports a false positive for the memory search check when using the local provider with node-llama-cpp.

Steps to reproduce

1. Set up local memory search with node-llama-cpp (provider: local)
2. Download a GGUF embedding model (e.g., embeddinggemma-300m-qat-Q8_0.gguf)
3. Verify it works: openclaw memory status --deep shows "Embeddings: ready"
4. Run openclaw doctor --non-interactive

Expected

Doctor should detect the working local model and report memory search as healthy.

Actual

Doctor reports:

Memory search provider is set to "local" but no local model file was found.

Fix (pick one):
- Install node-llama-cpp and set a local model path in config
- Switch to a remote provider: openclaw config set agents.defaults.memorySearch.provider openai

Meanwhile openclaw memory status --deep confirms everything is working:

Provider: local (requested: local)
Model: hf:ggml-org/embeddinggemma-300m-qat-q8_0-GGUF/embeddinggemma-300m-qat-Q8_0.gguf
Embeddings: ready
Indexed: 197/257 files · 2595 chunks
Embedding cache: enabled (7641 entries)

Root cause (likely)

Doctor appears to look for a model path in openclaw.json config rather than checking where node-llama-cpp actually stores downloaded models (~/.node-llama-cpp/models/). The config only has "provider": "local" without an explicit model path, but node-llama-cpp resolves the model from its own cache at runtime.

Environment

• OpenClaw: v2026.2.25
• Platform: Linux x86_64 (Ubuntu 24.04)
• node-llama-cpp models dir: ~/.node-llama-cpp/models/