Skip to content
DevAgentRadarDevAgentRadar
Sign in
Proposals/Switch out of Cline plan mode yourself after rev
proposaleveryoneP5High impactCline

Switch out of Cline plan mode yourself after reviewing the plan

Cline v4.1.4 removes model-initiated plan-to-act switching and hard-blocks file-editing shell commands in plan mode. Treat plan mode as read-only investigation, then explicitly toggle to Act to execute the presented plan.

Why this loop

The model can no longer leave plan mode mid-turn; you drive Plan→Act. Mutating shell use is refused in plan mode (file writes, in-place editors, redirects to files, mutating git, package installs) while read-only investigation still works. A turn that only hits a plan-blocked command and finishes with a plan no longer lands in a failed Retry state—toggling to Act re-runs that plan. Adopt a fixed loop: investigate in Plan, human-approve, manually switch to Act. Also use the slash menu where skills sit beside workflows with name disambiguation, and rely on clearer missing-API-key errors plus empty-response retries across providers.

Proposed actions

  1. Upgrade Cline to v4.1.4, start a task in Plan mode, and confirm the Plan/Act tooltip keybinding label matches your OS.
  2. In Plan mode, tell Cline: "Investigate only—list files and commands you would change; do not edit." Attempt (or have it attempt) a file write, `git commit`, and a package install; confirm each is hard-blocked.
  3. When the plan is shown, manually toggle Plan→Act yourself. Confirm the presented plan re-runs and the turn does not stick on a Retry footer or no-op.
  4. Open the slash command menu; verify skills appear alongside workflows and pick the disambiguated entry when two commands share a name.
  5. Run a short probe on your usual provider (OpenRouter/Cline/OpenAI-compatible/Bedrock/Claude 4.6+/5.x): confirm empty responses retry, missing API keys surface a clear error, and Claude is not rejected with "thinking.type.enabled is not supported".

Agent prompt

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

Paste into Claude Code / CLAUDE.md task

# DevAgentRadar → Claude Code

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

## Context
Assistant: Cline
Proposal: Switch out of Cline plan mode yourself after reviewing the plan
Summary: Cline v4.1.4 removes model-initiated plan-to-act switching and hard-blocks file-editing shell commands in plan mode. Treat plan mode as read-only investigation, then explicitly toggle to Act to execute the presented plan.
Primary source: https://github.com/cline/cline/releases/tag/v4.1.4

## Why it matters
The model can no longer leave plan mode mid-turn; you drive Plan→Act. Mutating shell use is refused in plan mode (file writes, in-place editors, redirects to files, mutating git, package installs) while read-only investigation still works. A turn that only hits a plan-blocked command and finishes with a plan no longer lands in a failed Retry state—toggling to Act re-runs that plan. Adopt a fixed loop: investigate in Plan, human-approve, manually switch to Act. Also use the slash menu where skills sit beside workflows with name disambiguation, and rely on clearer missing-API-key errors plus empty-response retries across providers.

## Suggested actions
1. Upgrade Cline to v4.1.4, start a task in Plan mode, and confirm the Plan/Act tooltip keybinding label matches your OS.
2. In Plan mode, tell Cline: "Investigate only—list files and commands you would change; do not edit." Attempt (or have it attempt) a file write, `git commit`, and a package install; confirm each is hard-blocked.
3. When the plan is shown, manually toggle Plan→Act yourself. Confirm the presented plan re-runs and the turn does not stick on a Retry footer or no-op.
4. Open the slash command menu; verify skills appear alongside workflows and pick the disambiguated entry when two commands share a name.
5. Run a short probe on your usual provider (OpenRouter/Cline/OpenAI-compatible/Bedrock/Claude 4.6+/5.x): confirm empty responses retry, missing API keys surface a clear error, and Claude is not rejected with "thinking.type.enabled is not supported".

## Config surfaces this release may change
- MCP servers — check your repo before applying


## 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.
mcpmodelclibreakingRelease source ↗

Your loop

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

Community signal · 0 approved · 0 deferred · 0 ignored

Anonymous public tally (one vote per visitor; changing choice updates your vote). Not moderated product truth.

Originating release signal

Clinev4.1.4Aug 5, 2026

v4.1.4

Recognize Chutes as a provider. · Show skills alongside workflows in the slash command menu, and disambiguate commands that share a name instead of letting one shadow the other. · Remove model-initiated plan-to-act switching. · Switching out of plan mode is now driven by you, not by the model deciding mid-turn. · Hard-block file-editing shell commands in plan mode instead of relying on prompting alone. · Read-only investigation still works, but file manipulation, in-place editors, redirection to files, mutating git subcommands, and package installs are refused. · +15 more changes

Verified excerpt — the source's own words
### Added

- Recognize Chutes as a provider.
- Show skills alongside workflows in the slash command menu, and disambiguate commands that share a name instead of letting one shadow the other.

### Changed

- Remove model-initiated plan-to-act switching. Switching out of plan mode is now driven by you, not by the model deciding mid-turn.
- Hard-block file-editing shell commands in plan mode instead of relying on prompting alone. Read-only investigation still works, but file manipulation, in-place editors, redirection to files, mutating git subcommands, and package installs are refused.

### Fixed

- Stop treating a turn that completes with a plan as a failed turn when a plan-blocked command was its only tool call. The turn no longer ends in the error state with a Retry footer, and toggling to Act correctly re-runs the presented plan instead of appearing to do nothing.
- Show tool paths relative to the workspace in the chat view instead of absolute paths.
- Reset pending attachments when starting a new task, so images from the previous task no longer carry over.
- Surface a clear error when the selected provider has no API key configured, instead of a generic failure.

Excerpt ends here — this release continues at the source ↗.

Primary source ↗