Cline agent
ComponentUpdates related to the agent component of Cline.
All AGENT Features
- Enabled ClinePass for all CLI users(cli-v3.0.35)
- Added support for Tencent TokenHub as a new provider(cli-v3.0.35)
- Refreshed the bundled model catalog with the latest provider models(cli-v3.0.35)
- Added the SDK-backed VS Code extension runtime, enabling Cline to run tasks through the shared Cline SDK session layer for agent turns, tools, Plan/Act mode coordination, and more.(v4.0.0)
- Introduced the ClinePass integration into the VS Code extension, including onboarding, provider selection, signup/subscription handoff, live model lists, and clearer authentication/error handling.(v4.0.0)
- Enabled the Customize marketplace within the extension for discovering and managing Skills, MCP servers, and Plugins, featuring installed/marketplace tabs, search/filtering, and install/uninstall flows.(v4.0.0)
- Launched Cline Plugins, allowing users to extend Cline with custom tools, workflows, and MCP-powered capabilities installed via the new Customize marketplace.(v4.0.0)
- Added queued prompts in chat, allowing messages submitted while Cline is busy to queue, display while streaming, and be cancelled before execution.(v4.0.0)
- Added edit-and-regenerate support for previous user messages, accompanied by clearer Reset Chat and Reset Code actions.(v4.0.0)
- Added generic SDK provider settings and model-catalog support, enabling multiple providers to share a model picker, reasoning controls, and dynamic model ID handling.(v4.0.0)
- Added support for numerous new and updated models/providers via the SDK, including Fireworks GLM 5.2, Kimi K2.6 Fast, Kimi K2.7 Code, Qwen 3.7 Plus, and SAP AI Core wiring.(v4.0.0)
- Added a new `cline skill` command for installing and managing skills, mirroring the functionality of `cline plugin install` and `cline mcp`.(cli-v3.0.27)
- Introduced a prefilled MCP install wizard command to streamline the setup process for MCP servers.(cli-v3.0.27)
- Added ClinePass support, with selectable ClinePass models available in the model picker(cli-v3.0.25)
- Added MCP server support to plugins, including authorizing plugin MCP OAuth during installation(cli-v3.0.25)
- Enabled plugin commands to submit prompts directly to the agent(cli-v3.0.24)
- Added support for overriding the API base URL configuration(cli-v3.0.24)
- Enabled a new --allowed-user-id flag for the Telegram connector to restrict which Telegram users can interact with the agent.(cli-v3.0.17)
- Added support for the GPT-5.5 model within SAP AI Core.(v3.85.0)
- Introduced support for DeepSeek V4 Flash and Pro models.(v3.85.0)
- Enabled Gemini 3.5 Flash model integration for Gemini and Vertex providers.(v3.85.0)
- Added support for the /lg-task URI webhook integration for LG dashboard flows.(v3.85.0)
- Enabled accepting dash-prefixed prompts when passed after --, preventing prompts starting with - from being parsed as flags.(cli-v3.0.9)
- Added the --worktree flag to automatically create a fresh git worktree under ~/.cline/worktrees/ and run the task there, supporting resumption with --taskId and --continue.(cli-v3.0.3)
- Enabled bearer token authentication for AWS Bedrock and improved AWS profile mapping during Bedrock gateway configuration.(cli-v3.0.3)
- Enabled enterprise-managed remote global skills to appear in a dedicated "Enterprise Skills" section with UI toggle support and system prompt integration.(v3.80.0)
- Introduced dynamic fetching of recommended models during the onboarding flow, with a fallback mechanism.(v3.80.0)
- Added a dedicated "Quota Exceeded" error message to the chat UI when account spend caps are reached.(v3.80.0)
- Added support for skills and optional modelId in subagent configuration(v3.67.0)
- Introduced AgentConfigLoader for loading agent configurations from files(v3.67.0)
- Enabled Responses API support for the OpenAI native provider(v3.67.0)
- Added /q command to quit the Command Line Interface (CLI)(v3.67.0)
- Added dynamic flag to adjust banner cache duration(v3.67.0)
- Introduced native `use_subagents` tool to replace legacy subagents.(v3.58.0)
- Enabled support for bundling `endpoints.json` so packaged distributions can ship required endpoints.(v3.58.0)
- Added support for parallel tool calling with Amazon Bedrock.(v3.58.0)
- Introduced a new experimental "double-check completion" feature to verify work before marking tasks complete.(v3.58.0)
- Added new task controls/flags to the CLI, including custom `--thinking` token budget and `--max-consecutive-mistakes` for yolo runs.(v3.58.0)
- Added new UI/options (including connection/test buttons) and support for syncing deletion of remotely configured MCP servers via remote config.(v3.58.0)
- Enabled 1M context model options for Claude Opus 4.6 in Vertex / Claude Code.(v3.58.0)
- Added support for ZAI/GLM: GLM-5.(v3.58.0)
- Introduced Cline CLI 2.0, installable via `npm install -g cline`(v3.57.0)
- Enabled access to Anthropic Opus 4.6 model(v3.57.0)
- Enabled access to Minimax-2.1 and Kimi-k2.5 models for a limited-time free promotion(v3.57.0)
- Enabled access to Codex-5.3 model for ChatGPT subscribers(v3.57.0)
- Added the Skills system for defining and using reusable, on-demand agent instructions.(v3.48.0)
- Introduced new websearch tooling within the Cline provider.(v3.48.0)
- Enabled support for the zai-glm-4.7 model in the Cerebras model list.(v3.48.0)
- Improved model refresh and reasoning support when interacting with the Vercel AI Gateway.(v3.48.0)
All AGENT Bug Fixes
- Fixed first-prompt truncation issue on high-output models (like MiniMax M3) where the initial task was being cut down to just the input wrapper due to premature auto-compaction(cli-v3.0.35)
- Fixed marketplace edge cases related to refreshing MCP servers after installs, remote configuration disabling of the MCP Marketplace tab, hiding workflows, surfacing plugin-bundled skills, and uninstalling shared marketplace entries.(v4.0.0)
- Fixed chat submission during active turns by queuing new messages instead of dropping or racing them, ensuring pending/queued states display promptly, and removing delayed send behavior.(v4.0.0)
- Fixed editing previous user messages, ensuring Escape cancels editing locally and clarifying reset action labels.(v4.0.0)
- Fixed terminal reliability issues, including standalone Windows output capture, hardened PowerShell command handling, correct running-state display, and preservation of raw structured commands.(v4.0.0)
- Fixed SDK tool-result and provider-message budgeting by implementing default truncation for large tool outputs, capping assistant text, limiting ingestion of bash/file-read/search output, and bounding media budgets.(v4.0.0)
- Fixed login and feature-flag resolution by ensuring the correct user/account identity is used on startup and simplifying the login user experience.(v4.0.0)
- Fixed issue where the CLI would silently treat unknown commands or unquoted multi-word input as a prompt; the CLI now rejects these inputs with a clear error message.(cli-v3.0.27)
- Fixed apply_patch to fail correctly when a hunk is skipped(cli-v3.0.25)
- Fixed run_commands to return captured stdout on failure and correctly handle split heredocs(cli-v3.0.25)
- Fixed search tools to treat zero results as a successful operation(cli-v3.0.25)
- Fixed disabled-reasoning handling specifically for StepFun flash operations(cli-v3.0.25)
- Fixed history resume rendering isolation issues(cli-v3.0.25)
- Fixed incorrect Cline OAuth token formatting within the provider configuration(cli-v3.0.25)
- Fixed truncation of structured tool operation result strings, ensuring oversized tool output is properly contained(cli-v3.0.24)
- Stopped echoing the full command text within run_commands tool results for cleaner output(cli-v3.0.24)
- Fixed issue where Slack channel mentions did not post replies in the original message's thread.(cli-v3.0.18)
- Fixed the abort indicator so it clears immediately upon task cancellation.(cli-v3.0.18)
- Fixed an issue where the interactive CLI would get stuck after stopping and restarting Cline Hub and pressing Escape to cancel a request; the CLI now recovers pending messages and starts a fresh session.(cli-v3.0.17)
- Resolved race conditions between Ctrl+C and Hub shutdown that caused "hook dispatch failed" and WebSocket connection errors.(cli-v3.0.17)
- Fixed the Hub daemon shutting down prematurely when a runtime request was aborted, ensuring the daemon remains active.(cli-v3.0.17)
- Fixed incorrect global endpoint handling for Claude models when using Vertex AI.(v3.85.0)
- Ensured Poolside Laguna models are correctly routed through next-gen prompts and native tool calling.(v3.85.0)
- Fixed runtime host crash that occurred after a cancel operation hit an error during hub abort cleanup.(cli-v3.0.9)
- Fixed GLM thinking not behaving correctly through the gateway by routing it through provider metadata.(cli-v3.0.9)
- Fixed dropped macOS screenshots when pasting or reading them by correctly resolving paths containing Unicode variants like U+202F (narrow no-break space).(cli-v3.0.3)
- Resolved a crash that occurred when encountering detached hub event errors during a session.(cli-v3.0.3)
- Ensured that Ollama models with reasoning enabled by default now correctly honor the --thinking none flag.(cli-v3.0.3)
- Fixed Out-Of-Memory (OOM) crashes during long conversations by increasing the Node.js process memory allocation to 8GB.(v3.80.0)
- Resolved issue where generic error messages were shown in the chat error row; now displays detailed error information.(v3.80.0)
- Fixed a crash related to reasoning delta when processing usage-only stream chunks(v3.67.0)
- Corrected an issue where OpenAI tool ID transformation was incorrectly restricted to the native provider only(v3.67.0)
- Fixed authentication check failure when operating in ACP mode(v3.67.0)
- Resolved an issue where the CLI yolo setting was being incorrectly persisted to disk(v3.67.0)
- Fixed the inline focus-chain slider to correctly operate within its feature row(v3.67.0)
- Ensured compatibility with Gemini 3.1 Pro(v3.67.0)
- Corrected authentication issues when using the Cline CLI with the ACP flag(v3.67.0)
- Fixed CLI to correctly handle stdin redirection in CI/headless environments.(v3.58.0)
- Fixed CLI to preserve OAuth callback paths during auth redirects.(v3.58.0)
- Fixed OAuth callback reliability in VS Code Web by generating callback URLs via `vscode.env.asExternalUri`.(v3.58.0)
- Fixed Terminal to surface command exit codes in results and improved long-running `execute_command` timeout behavior.(v3.58.0)
- Fixed UI rendering by adding a loading indicator and fixing `api_req_started` rendering.(v3.58.0)
- Prevented duplicate streamed text rows after task completion during task streaming.(v3.58.0)
- Fixed API to preserve the selected Vercel model when model metadata is missing.(v3.58.0)
- Ensured telemetry flushes on shutdown and routed PostHog networking through proxy-aware shared fetch.(v3.58.0)
- Increased Windows E2E test timeout to reduce flakiness in CI.(v3.58.0)
- Fixed issue where the read file tool failed to support reading large files(v3.57.0)
- Resolved crash occurring when entering decimal input in OpenAI Compatible price fields (#8129)(v3.57.0)
- Corrected build complete handlers when updating the API configuration(v3.57.0)
- Fixed missing provider from the model list(v3.57.0)
- Fixed issue where the Favorite Icon/Star was getting clipped in the task history view(v3.57.0)
- Fixed an extension crash that occurred when using the context menu selector.(v3.48.0)
- Reverted a previous change (#8341) that caused regressions in the diff view and document truncation.(v3.48.0)
Releases with AGENT Changes
sdk/sdk/v0.0.63cli-v3.0.353 features1 fixThis 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.
v4.0.08 features6 fixesThis major release introduces the SDK-backed VS Code extension runtime, unifying agent coordination, tool execution, and configuration management under a shared SDK session layer. Key new user capabilities include the Customize marketplace for managing Plugins and Skills, and queued prompts in chat. Numerous fixes address stability in terminal execution and SDK budgeting, alongside significant architectural changes for improved provider and model configuration consistency.
cli-v3.0.272 features1 fixThis release introduces enhanced command-line management with the new `cline skill` command for managing skills and a streamlined wizard command for quicker MCP server setup. Error handling has also been improved, particularly for OAuth authorization failures, and the CLI now provides clearer feedback for invalid commands.
cli-v3.0.252 features6 fixesThis release introduces ClinePass model support and enhanced MCP server integration for plugins. Key improvements include better handling of large file reads and capped tool output to manage context limits, alongside several fixes for patching, command execution, and search tools.
cli-v3.0.242 features2 fixesThis release introduces the ability for plugin commands to submit prompts to the agent and adds configuration support for overriding the API base URL. Key fixes include resolving issues with truncated tool output and suppressing flickering console windows on Windows.
cli-v3.0.182 fixesThis release focuses on improving integration reliability and ensuring the local environment uses the latest dependencies. Key fixes include ensuring Slack replies thread correctly and clearing task indicators immediately upon cancellation.
cli-v3.0.171 feature3 fixesThis release focuses heavily on stability, particularly around the interactive CLI and Hub daemon lifecycle. Key fixes address session recovery in the CLI and prevent premature daemon shutdowns during request abortions. Additionally, the Telegram connector now supports user authorization via a new flag.
v3.85.04 features2 fixesThis release significantly expands model compatibility by adding support for GPT-5.5, DeepSeek V4, and Gemini 3.5 Flash across various providers. It also introduces a new webhook integration for LG dashboard flows and resolves an endpoint handling issue for Claude models on Vertex AI.
cli-v3.0.91 feature2 fixesThis release focuses heavily on performance and stability improvements, including faster CLI startup and configuration toggles. Key user experience updates include restoring fuzzy ranking for file mentions and ensuring interactive sessions remain active after task cancellation. Additionally, prompts starting with a dash are now correctly accepted.
cli-v3.0.32 features3 fixesThis release introduces isolated task execution via the new --worktree flag, allowing users to resume experiments in clean environments. Key fixes include resolving issues with dropped macOS screenshots due to Unicode path problems. Additionally, the system prompt has been refined for better guidance on tool usage and verification steps.
v3.80.03 features2 fixesThis release introduces significant enterprise integration by enabling remote, managed global skills with full UI support. It also improves stability by fixing Out-Of-Memory crashes during long sessions and enhances user feedback with more detailed error messages in the chat interface. Furthermore, task execution has been streamlined by defaulting all commands to background mode, removing the dependency on the VS Code integrated terminal.
v3.67.05 features7 fixesThis release introduces significant configuration flexibility by adding support for skills and file-based agent configurations via the new AgentConfigLoader. Key fixes address stability issues, including a reasoning delta crash and compatibility problems with Gemini 3.1 Pro. Users will also benefit from reduced response latency due to websocket preconnection.
v3.58.08 features9 fixesThis release introduces significant new capabilities, including native subagent support, parallel tool calling for Amazon Bedrock, and an experimental double-check feature for task verification. Several critical bugs were fixed across the CLI, VS Code Web, and Terminal, alongside UX improvements like moving reasoning effort into model settings.
v3.57.04 features5 fixesThis release introduces Cline CLI 2.0 and expands model availability, adding Anthropic Opus 4.6, Minimax-2.1, Kimi-k2.5, and Codex-5.3. Several critical bugs were fixed, including issues with reading large files and crashes in price input fields, alongside a UX change making skills permanently enabled.
v3.48.04 features2 fixesThis release introduces the powerful new Skills system for creating reusable agent instructions and adds new websearch tooling to the Cline provider. We also addressed a critical bug causing extension crashes when using the context menu selector and reverted a recent change that introduced regressions in the diff view.