Continue Ollama config hides local models
This looks more like a Continue YAML-schema mismatch than an Ollama problem. Continue's current docs expect schema v1 and fold autocomplete into the main models list, so a config that looks close can still load as "No models configured."
The likely failure mode is that the extension is parsing the file as invalid or legacy config, then falling back to an empty model registry.
- –Continue’s YAML docs require `schema: v1`; the posted config shows `version: 1` but no schema header, which is the biggest red flag.
- –In YAML, `tabAutocompleteModel` is deprecated; autocomplete should live under `models` with `roles: [autocomplete]`.
- –`apiBase: http://127.0.0.1:11434` matches the Ollama docs, so the transport layer looks fine compared with the config shape.
- –On Windows, it is worth verifying the file is actually in `%USERPROFILE%\.continue\config.yaml`, not just a path that looks right in another shell.
- –If the user just wants local Ollama detection, Continue’s `model: AUTODETECT` path is the docs-backed fallback that populates from local `ollama list`.
DISCOVERED
119d ago
2026-04-01
PUBLISHED
119d ago
2026-04-01
RELEVANCE
AUTHOR
Existing-Monitor-879