Force a default provider/model & disable “sticky models”

[Mgrmjp]

Feature request: Force a default provider/model & disable “sticky models”

Summary

I need a way to always use a specific Configuration Profile and Model in Kilo Code, and to disable any sticky/persistent model behavior so the extension never auto-switches to a previously used provider/model.

Currently I can select a profile (e.g., “GLM4.6”) in the Providers UI, but Kilo sometimes reverts or sticks to another model used earlier.
There doesn’t appear to be a setting to force a default profile/model or to disable stickiness.

---

Why this matters

• Repeatable builds & CI-like runs — teams want deterministic behavior per workspace/repo.
• Multiple backends in rotation (OpenAI-compatible, Claude-compatible, local): the wrong default causes capability/cost drift and confusing diffs.
• Enterprise policy — some repos must only talk to a specific base URL/provider.

---

Expected behavior

• On window reload / extension activation / new “Ask”:
  • Kilo Code uses my configured default profile and my configured default model.
  • “Sticky” session memory does not override these defaults unless I explicitly choose a different model for that request.

---

Actual behavior

• No visible setting to set a default profile/model.
• No visible setting to disable “sticky models”.
• Extension sometimes keeps using whatever model was last active in another file/session.

---

Proposed solution

Introduce three settings (workspace & user scope):

// Force which “Configuration Profile” to load on activation.
"kilo.defaultConfigurationProfile": "GLM4.6",

// Force the initial model for that profile on activation.
"kilo.defaultModel": "glm-4.6",

// If false, do not persist last-used model across files/sessions.
"kilo.stickyModels": false