Cline

This release introduces Agent Plugins managed through the Cline Hub, allowing for easier discovery and execution of custom tools. Sign-in processes are improved with in-app device confirmation codes, and voice input failures now provide clearer guidance to settings.

This release introduces image generation capabilities, allowing users to create images with server-side credential management and session history access. It also enhances the organization of scheduled runs by grouping them within their runtime environment and clarifies media-generation settings for SSH environments.

This release enhances the Cline experience by integrating ClinePass information throughout the app and improving the reliability of various features. Key fixes include addressing memory leaks, preventing crashes related to hook scripts and API requests, and clarifying error messages for API key issues and sign-in problems.

This release introduces the ability to import chat history from other tools and enables voice input on macOS. Several bugs have been fixed, including issues with tool calling and message editing, and improvements have been made to session organization and the marketplace detail panel.

This release enhances the CLI's interaction with older Hub versions by introducing a clear warning prompt before replacement. It also fixes several critical bugs, including issues with tool calling, image handling, and Langfuse tracing in released builds. Additionally, the model catalog has been significantly refreshed with new providers and updated model defaults.

This release introduces a new service for importing conversation history from various sources like Claude Code and Codex, making it easier to bring existing data into Cline. It also enhances schedule management by anchoring them to a central location and allowing cross-workspace access, while improving the searchability of session history through a server-side index.

This release enables web search by default for new desktop sessions. It also includes several improvements to Composio connectors for better resilience and introduces a more robust Marketplace explorer with reliable agent cancellation and clearer authentication errors.

This release overhauls the Marketplace with a new two-pane explorer and introduces the ability to drop file attachments anywhere in the chat. It also ensures that stopping sessions now fully terminates all related processes and improves how authentication errors are handled for provider models.

This beta release introduces powerful cloud integration, allowing you to seamlessly hand off local sessions to Cline Cloud and recover interrupted work. You can now also choose your preferred coding environment—local, SSH remote, or Cloud—directly from the desktop app.

This release introduces inline image rendering for tool results and a more robust session search covering your full history. Key improvements include a new onboarding step for GitHub integration and enhanced Windows shell stability, alongside numerous bug fixes for scheduled tasks, sign-ins, and workspace management.

This release introduces seven new AI providers to the model catalog and enhances security by redacting git credentials. Several bugs have been fixed, including crashes in the codebase search and issues with MCP server installations, along with memory leaks in the background hub process.

This release significantly reduces memory usage and event payload sizes by optimizing how session transcripts are handled. Session snapshots are now state-only, with transcripts fetched on demand, preventing the flooding of event logs and excessive memory consumption.

This release significantly enhances the model catalog by adding seven new providers and updating existing ones. It also resolves a critical bug where the Cline process could consume excessive memory during long sessions, ensuring a more stable experience.

This release enhances the `tasks` tool by allowing configuration for scheduled work only and improves security by redacting git credentials from system prompts. Several bugs have been fixed, including crashes in `search_codebase` and incorrect argument parsing in `installMcpServer`, alongside general improvements to file handling and model catalog updates.

This release focuses on stability and correctness, addressing several bugs related to cost estimation, provider selection, and tool performance. It also clarifies MCP auto-approval settings by removing redundant per-tool checkboxes.

This release introduces a new dedicated Marketplace for discovering plugins and servers, and reorganizes the sidebar for better session management. Several bugs have been fixed, including installation issues and display problems with scheduled runs, alongside improvements to the user interface and data handling.

This release introduces a unified 'Customize' hub for managing plugins and tools, and redesigns the Models page for better provider management. You'll also find improvements to session organization, voice input settings, and search functionality, alongside several important bug fixes.

This release updates the model catalog with new providers and pricing information, and refreshes default model choices for some services. It also addresses an issue with an outdated promotional offer in the initial setup dialog and optimizes the hub's event log to prevent excessive disk usage.

This release refreshes the model catalog with new providers and updated pricing. It also addresses a critical bug where task completion telemetry was dropped and improves the management of the hub's event log to prevent excessive disk usage.

This release addresses a bug related to the "Use MCP servers" toggle. The toggle now correctly auto-approves all MCP tool calls when enabled, ensuring it functions as intended without requiring individual tool opt-ins.

This release refreshes the built-in model catalog with a variety of new AI models, including Claude Fable 5 and several DeepSeek V4 Flash variants. It also resolves a bug that prevented task completion telemetry from being reported in interactive sessions.

This release introduces robust hub management with new drain and upgrade commands, ensuring continuous availability. Sessions are now resilient to hub restarts, and a critical bug preventing tool usage with custom OpenAI-Compatible models has been fixed.

This release introduces significant improvements to hub reliability, allowing for draining and upgrading without data loss and ensuring durable runs. Additionally, tool calling issues with custom OpenAI-Compatible models have been resolved, and Langfuse traces now provide better session and client identity tracking.

This release introduces robust agent handoff between Hub instances, preventing work loss during restarts. It also ensures the app respects server-side feature flags and fixes an issue with tool calling for custom OpenAI-Compatible models.

This release addresses several critical bugs. Tool calling for custom OpenAI-Compatible models has been restored, and Hub-backed sessions are now more resilient to restarts and upgrades. Additionally, session and client identity are correctly carried over to Langfuse traces.

This release focuses on bug fixes related to enterprise controls and custom model tool calling. MCP controls are now correctly enforced on the Customize marketplace, and tool calling has been restored for custom OpenAI-Compatible models with empty capability lists.

This release enhances the user experience with a redesigned onboarding flow and improved prompt handling during cloud handoffs. Several critical bugs have been fixed, including issues with prompt loss, popovers blocking clicks, and checkpoint restore functionality.

This release rebrands the app to Cline, introduces a unified Plugins hub with a marketplace, and enhances agent capabilities with durable todos and scheduling. Users will also experience a streamlined model selection interface and improved provider support.

This release introduces the ability for models to generate images directly within tasks, rendering them inline in your conversation. Several bugs have been fixed, including issues with code actions, file path handling, and task session management. Additionally, the model catalog has been refreshed with new providers and updated information.

This release enhances media generation capabilities, allowing models to produce images directly in conversations and improving how skill slash commands are handled. Several bugs have been fixed, including issues with dialog theming, tool activity display, and command execution.

This release introduces client-scoping for the `tasks` tool, ensuring better reliability by only allowing capable clients to register and use tasks. The tool catalog resolution has also been improved to accurately reflect tool availability based on client types across different sessions.

This release introduces model-driven image generation and allows agents to manage scheduled tasks and a todo agenda. Several bugs have been fixed, including issues with tool activity, command execution, and session status reporting, alongside improvements to model provider information and command output streaming.

This beta release introduces the ability to hand off local sessions to Cline Cloud, allowing your work to continue even after closing the app. Handoffs are now more robust with recovery options for interruptions, and cloud sessions are integrated into the main environment menu.

This release introduces native macOS notifications and voice input for enhanced background operation and dictation. Key improvements include streaming command output, inline image generation, and a redesigned agent decision card for a more interactive experience. Several bug fixes address issues with command execution, session prompts, and custom base URL support.

This release focuses on the backend infrastructure for beta desktop app auto-updates. It introduces a rolling release model for beta versions, ensuring that beta installs automatically poll for and receive the latest updates.

This first beta release introduces powerful new features like cloud sessions and an avatar overlay, allowing you to run and sync your coding sessions in the cloud and get visual feedback on your desktop. It also includes significant improvements from the upcoming stable release, such as voice input and model-driven image generation.

This release introduces significant new capabilities, including adjustable app font size and web search integration for supported models. It also addresses several critical bugs, such as issues with extended thinking budgets and conflicting Cline installations, ensuring a more stable and user-friendly experience.

This release introduces web search capabilities for supported models, allowing them to search the web during tasks. Several bugs have been fixed, including issues with daemon shutdowns, Hub replacements, and plugin sandbox processes. The model catalog has also been updated with a new provider and refreshed model lists.

This release enhances update stability by ensuring auto-updates install only when no CLI is attached, preventing session interruptions. It also improves the streaming assistant's markdown rendering to prevent flashing and jumping, and introduces web search result visibility directly in the transcript.

This release introduces provider-executed web search, allowing models to access the internet during conversations. It also enhances the Cline gateway with a dedicated provider and fixes several critical bugs related to Hub daemon stability and session handling.

This release enhances the Claude Code provider for agentic tasks, improving its ability to run native tools and manage permissions. Several critical bugs have been fixed, including issues with token refreshing, telemetry spam, and tool call processing, ensuring a more stable and reliable experience.

This release significantly improves the Claude Code provider, fixing issues with tool execution and session anchoring. Several critical bugs have been resolved, including problems with tool-call JSON repair, strict provider rejections, and mid-turn crashes during streamed tool calls. Additionally, the Hub daemon upgrade process has been refined to prevent conflicts between different Cline installations.

This release enhances the Claude Code provider for agentic work, improving tool handling and session management. Several critical bugs have been fixed, including issues with tool-call JSON, provider rejections, and mid-turn crashes. Telemetry reporting has also been optimized for better accuracy and reduced volume.

This release overhauls the transcript experience with individual tool call rows, terminal-like command display, and inline diffs for edits. Key bug fixes include improved daemon handling and corrected behavior for strict providers, while enhancements focus on typography, UI layout, and more accurate token usage for model reasoning.

This release introduces Fable 5 to the Vertex model catalog and ensures custom Vertex model IDs are routed correctly. Several bugs have been fixed, including issues with CLI daemon reconnections, compaction on reasoning models, and handling of empty summaries.

This release introduces Fable 5 to the Vertex model catalog and improves custom model ID routing. Key bug fixes include resolving stale daemon reconnections and ensuring model compaction is no longer skipped, with the summarizer now respecting explicit token limits.

This release enhances Vertex AI by allowing manual entry of any model ID and adding support for Fable 5. It also improves the model catalog display and streamlines unattended run configurations by consolidating approval settings.

This release introduces several new features, including the ability to paste images directly into the composer and improved handling of Git repositories and folder content. Bug fixes address issues with diff panel path resolution, failed turns, and startup errors, while performance and UX improvements enhance the overall user experience.

This release brings back the "View Changes" button with SDK checkpoint support, allowing detailed review of task modifications. It also reintroduces a copy button for responses and enhances session and prompt handling during interruptions and failures, ensuring greater data preservation and reliability.

This release enhances session management by ensuring context durability across restarts and interruptions, and improves the handling of queued prompts. It also introduces a new command for uninstalling MCP servers and fixes several rendering and network-related issues for a more robust user experience.

This release enhances session durability, ensuring prompts and context survive interruptions and restarts. It also improves error reporting for failed turns and network issues, while refining how LiteLLM requests and remote MCP servers are handled.

This release introduces OAuth authentication for remote MCP servers and improves error handling for failed turns and invalid server configurations. Users will also benefit from a more intuitive welcome screen with a "Connect a model" prompt and enhanced keyboard shortcuts for improved workflow.

This release introduces a new model, `meta/muse-spark-1.2-contributor`, to the Cline provider, expanding the available model catalog. Additionally, a bug has been fixed to ensure telemetry accurately reflects the model used for a run, improving error reporting.

This release ensures reasoning effort is applied consistently across all providers and respects user preferences for disabling reasoning. You can now select the `meta/muse-spark-1.2-contributor` model on the Cline provider, and the model catalog has been refreshed.

This release introduces a more unified and portable approach to reasoning settings across different AI SDK providers, including Ollama. It also improves telemetry accuracy by correctly attributing errors to the model in use and stripping undefined values.

This release enhances the user experience when interacting with retired free models by providing clear notices and alternative model suggestions. Reasoning settings are now more consistent across various AI SDK providers, ensuring predictable behavior.

This release introduces Chutes as a new provider and enhances the slash command menu to show skills alongside workflows. It also brings significant improvements to plan-to-act switching, making it user-driven, and hard-blocks file-editing shell commands in plan mode for better security.

This release introduces a universal macOS download for native Apple Silicon and Intel support, along with enhanced session history filtering and a visual token usage ring in the composer. Key bug fixes include resolving empty model response failures across multiple providers and restoring Bedrock prompt caching.

This release introduces user-selectable color themes for the interactive TUI, allowing for a more personalized experience. It also enhances stability by fixing several bugs, including issues with git branch updates, Telegram commands, and connector launch collisions, while improving error handling for model responses and context management.

This release enhances Plan mode by preventing accidental file modifications and improves context window handling with automatic recovery. Sessions now track their origin and use lazy persistence, while empty responses are retried across all providers for greater reliability.

This release introduces the ability to select free Cline models via the ClinePass provider. Key fixes include ensuring accurate diff reporting for str_replace edits and preserving canonical session history during context compaction. Several minor provider and telemetry updates have also been implemented.

This release focuses on improving the ClinePass experience by adding clearer error handling for limit-reached scenarios and enabling users to easily select free Cline models. Additionally, onboarding copy and link handling have been refined for better usability.

This release introduces a significant visual refresh with a new plan/act accent color palette and a restyled chat input for a cleaner look. Key functional updates include improved handling of malformed input from weaker models and setting a default context window for models that don't specify one.

This release significantly improves reliability when using weaker models by gracefully handling their malformed tool calls. Additionally, the system now correctly communicates plan/act mode switches to the model mid-session and fixes an issue causing unexpected restarts when switching modes in empty sessions.

This release significantly improves mode switching reliability, ensuring that when you switch from plan mode to act mode, the full act-mode toolset is immediately available. We also fixed a bug that could cause unapproved plans to start execution due to timing issues with mode toggles.

This release enables ClinePass for all CLI users and introduces Tencent TokenHub as a new supported provider. Key fixes include resolving an issue where high-output models truncated the first prompt due to premature compaction.