Local AI diagnostics report Ollama as not running and required models as missing

[Al629176]

Summary

Local AI diagnostics report Ollama as not running and required models as missing.

Problem

In Settings > Local AI diagnostics, running diagnostics shows an Ollama error. The server is reported as not running, the binary is not found, and expected chat/embedding models are missing.

Expected behavior is that diagnostics should either detect and start/repair Ollama correctly, or provide clear setup actions for installing Ollama and downloading required models.

Steps to reproduce:

1. Open OpenHuman desktop app.
2. Go to Settings > Local AI / diagnostics.
3. Click Run Diagnostics.
4. Observe Ollama server not running or unreachable at localhost:11434 and required models marked missing.

Environment:

• Desktop app
• Settings > Local AI diagnostics
• Backend: Ollama
• Expected endpoint: http://localhost:11434
• Platform/version unknown

Solution optional

Check Ollama binary detection, configured binary path, bootstrap/resume flow, and model download checks. Improve diagnostics to show one-click repair actions for installing/starting Ollama and downloading missing models.

Acceptance criteria

• Repro gone - Diagnostics no longer report false Ollama not running/binary missing errors when Ollama is installed and reachable.
• Regression safety - Unit, integration, or E2E coverage added or updated for local AI diagnostics and bootstrap behavior.
• Diff coverage >= 80% - Fix PR meets changed-lines coverage gate.
• Clear repair flow - If Ollama is genuinely missing or stopped, UI provides clear install/start/download actions.
• Model detection - Required chat and embedding models show accurate installed/missing status.
• No dead-end state - Bootstrap/resume or force re-bootstrap can recover the local AI setup without manual terminal work where possible.

Related

Screenshot provided showing Ollama Diagnostics with server not running, binary not found, and missing gemma3:1b-it-qat / all-minilm:latest models.