Skip to content
DevAgentRadarDevAgentRadar
Sign in
Proposals/Re-select pinned Vultr models and page MCP overf
proposaldeveloperP5Worth a lookCline

Re-select pinned Vultr models and page MCP overflow with read_files

Cline CLI v3.0.67 lets the agent page MCP tool output past the context cutoff with read_files, actually runs custom providers.json/models.json entries, surfaces credential-save errors, and requires re-selecting pinned Vultr models after upstream ID renames.

Why this loop

cli-v3.0.67 closes three workflow holes. MCP tools that overflowed context used to drop the tail; the agent now gets a preview plus a pageable link via read_files, so long tool dumps stay recoverable—tell the agent to follow that link instead of treating the truncated preview as complete. Custom providers already appeared in pickers but failed at run time with Unknown or disabled provider; they now execute, so you can stop falling back to built-in providers for local or custom endpoints. Credential save failures during onboarding or the provider settings dialog now render in place instead of failing silently. The catalog refresh adds GPT-6.1 Sol to Cline recommended 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). Vultr IDs were renamed upstream, which is a breaking pin: any saved Vultr model must be re-selected or tasks will hit a stale id.

Proposed actions

  1. Upgrade Cline CLI to cli-v3.0.67, then open the model picker and re-select any pinned Vultr model (upstream IDs were renamed; the new default is MiMo V2.6 Flash RL).
  2. When an MCP tool returns a preview plus a link because output exceeded context, immediately call read_files on that link and page through the rest instead of treating the truncated preview as complete.
  3. Retry any task that previously failed with Unknown or disabled provider against a custom entry in providers.json/models.json; those providers now work when a task actually runs.
  4. If you use 302.AI, NanoGPT, Vivgrid, or Vultr, confirm the new defaults (Claude Sonnet 5.5, Ling 3.1 Flash, GPT-6.1 Sol, MiMo V2.6 Flash RL) and switch explicitly if you still want the previous model.
  5. If provider credentials fail to save during onboarding or in the provider settings dialog, read the in-place error (it no longer fails silently) and correct the credential before continuing.

Agent prompt

Paste into your agent or query via MCP (get_agent_prompt) — free, no extra AI cost

Cline / .clinerules task

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-select pinned Vultr models and page MCP overflow with read_files Summary: Cline CLI v3.0.67 lets the agent page MCP tool output past the context cutoff with read_files, actually runs custom providers.json/models.json entries, surfaces credential-save errors, and requires re-selecting pinned Vultr models after upstream ID renames. Primary source: https://github.com/cline/cline/releases/tag/cli-v3.0.67

Why it matters

cli-v3.0.67 closes three workflow holes. MCP tools that overflowed context used to drop the tail; the agent now gets a preview plus a pageable link via read_files, so long tool dumps stay recoverable—tell the agent to follow that link instead of treating the truncated preview as complete. Custom providers already appeared in pickers but failed at run time with Unknown or disabled provider; they now execute, so you can stop falling back to built-in providers for local or custom endpoints. Credential save failures during onboarding or the provider settings dialog now render in place instead of failing silently. The catalog refresh adds GPT-6.1 Sol to Cline recommended 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). Vultr IDs were renamed upstream, which is a breaking pin: any saved Vultr model must be re-selected or tasks will hit a stale id.

Suggested actions

  1. Upgrade Cline CLI to cli-v3.0.67, then open the model picker and re-select any pinned Vultr model (upstream IDs were renamed; the new default is MiMo V2.6 Flash RL).
  2. When an MCP tool returns a preview plus a link because output exceeded context, immediately call read_files on that link and page through the rest instead of treating the truncated preview as complete.
  3. Retry any task that previously failed with Unknown or disabled provider against a custom entry in providers.json/models.json; those providers now work when a task actually runs.
  4. If you use 302.AI, NanoGPT, Vivgrid, or Vultr, confirm the new defaults (Claude Sonnet 5.5, Ling 3.1 Flash, GPT-6.1 Sol, MiMo V2.6 Flash RL) and switch explicitly if you still want the previous model.
  5. If provider credentials fail to save during onboarding or in the provider settings dialog, read the in-place error (it no longer fails silently) and correct the credential before continuing.

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-cli-v3-0-67-re-select-pinned-vultr-models-and-page-mcp-overflow-wi and mark Applied, Skipped, or Failed. Proposal id: 3761803c-d49a-46cc-b7e8-afa291b2190c

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.

agentmcpmodelclibreakingRelease source ↗

Your loop

This browser · no sign-in · not shared as “you”

After you run the prompt

Only you can mark this. Agents cannot write your loop.

Your decision stays on this device. A public tally appears after a few votes.

Originating release signal

ClineCLI 3.0.67Sep 30, 2026

cli-v3.0.67 CLI v3.0.67

When an MCP tool returns more output than fits in context, the agent can now read the rest. · It gets a preview plus a link it can page through with read_files, so the output past the cutoff is no longer lost · Custom providers defined in providers.json/models.json now work when you run a task. · They showed up in the provider and model pickers but failed with Unknown or disabled provider · If saving provider credentials fails during onboarding or in the provider settings dialog, the error now shows in place instead of failing silently · +5 more changes
Verified excerpt — the source's own words
  • When an MCP tool returns more output than fits in context, the agent can now read the rest. It gets a preview plus a link it can page through with read_files, so the output past the cutoff is no longer lost
  • Custom providers defined in providers.json/models.json now work when you run a task. They showed up in the provider and model pickers but failed with Unknown or disabled provider
  • If saving provider credentials fails during onboarding or in the provider settings dialog, the error now shows in place instead of failing silently
  • On Linux, the auto-approve indicator in the status bar uses a glyph that common monospace fonts include, instead of rendering as a blank box or nothing
  • Refreshed the model catalog. The Cline recommended list adds GPT-6.1 Sol. Default models change for 302.AI (Claude Sonnet 5.5), NanoGPT (Ling 3.1 Flash), Vivgrid (GPT-6.1 Sol), and Vultr (MiMo V2.6 Flash RL). Vultr's model ids were renamed upstream, so a pinned Vultr model may need to be re-selected

Full Changelog: https://github.com/cline/cline/compare/cli-v3.0.66...cli-v3.0.67

Primary source ↗