DevAgentRadar → Cline

You are helping me adopt a real coding-assistant change. Work only from the facts below. Do not invent features.

Context

Assistant: Cline Proposal: Re-pin renamed Vultr models and retry Add Provider tasks in Cline Desktop v0.0.40 Summary: Cline Desktop v0.0.40 makes Add Provider custom backends runnable, remembers Cloud vs Local, unifies MCP settings paths, pages oversized MCP tool output, and refreshes the model catalog. Re-select pinned Vultr models after upstream id renames and retry previously failing custom providers. Primary source: https://github.com/cline/cline/releases/tag/desktop-v0.0.40

Why it matters

desktop-v0.0.40 unblocks custom providers added with Add Provider: they appeared in pickers but failed at run time with Unknown or disabled provider. Credential saves no longer fail when a model list cannot be fetched, and changing an API key or endpoint refreshes that list. New chats restore last Cloud vs Local mode and the last Cloud model; switching a thread from Local to Cloud also opens on that remembered Cloud model. MCP settings are now one file: with CLINE_MCP_SETTINGS_PATH, CLINE_DATA_DIR, or CLINE_DIR set, server read/write, MCP OAuth sign-in, and Open MCP settings share that path instead of splitting across files. When an MCP tool returns more than fits in context, the agent gets a preview plus a paging link so cutoff output is not lost. The catalog refresh is breaking for Vultr (upstream model ids renamed; pinned models may need re-selection) and changes defaults for 302.AI (Claude Sonnet 5.5), NanoGPT (Ling 3.1 Flash), Vivgrid (GPT-6.1 Sol), and Vultr (MiMo V2.6 Flash RL); GPT-6.1 Sol is on the Cline recommended list. Tool diffs now follow the app font size.

Suggested actions

1. In Cline Desktop v0.0.40, open the Vultr model picker and re-select your model — upstream ids were renamed, so a pinned Vultr model may no longer match. If you use 302.AI, NanoGPT, Vivgrid, or Vultr without a pin, expect the new defaults Claude Sonnet 5.5, Ling 3.1 Flash, GPT-6.1 Sol, and MiMo V2.6 Flash RL. Add GPT-6.1 Sol from the Cline recommended list if you want that catalog entry.
2. Select a custom provider created with Add Provider and run a task. If it previously failed with Unknown or disabled provider, it should now execute. After changing that provider's API key or endpoint, save credentials even if the model list cannot be fetched, then reopen the model picker so the list refreshes.
3. Point every Cline process at one MCP settings file: export CLINE_MCP_SETTINGS_PATH=/absolute/path/to/mcp_settings.json (or set CLINE_DATA_DIR / CLINE_DIR). Then use Open MCP settings, complete MCP OAuth sign-in, and save/load servers so all three hit that same path.
4. Instruct the Cline agent: when an MCP tool result exceeds the context window, follow the preview plus paging link to read the remainder; do not treat the cutoff as the full result or re-invoke the tool.
5. Start a new Cline chat without switching Cloud/Local: v0.0.40 restores your last Cloud vs Local choice and last Cloud model. When you switch an existing thread from Local to Cloud, confirm it opens on that remembered Cloud model instead of Local.

Config surfaces this release may change

• MCP servers — check your repo before applying

After you finish

Do not report this as applied to DevAgentRadar. You cannot write the visitor's loop.

Tell the human: open https://devagentradar.com/proposals/cline-desktop-v0-0-40-re-pin-renamed-vultr-models-and-retry-add-provider and mark Applied, Skipped, or Failed. Proposal id: 6603ecf0-4c4f-4a22-93e3-8c1bc082a2b1

Your job

1. Restate the change in one sentence.
2. Propose a minimal plan for my repo (or a throwaway pilot).
3. Implement only what I approve; prefer small diffs and tests.
4. Call out risks (permissions, breaking APIs, cost).

Start by confirming you understood the proposal.