Skip to content
| Marketplace
Sign in
Visual Studio Code>Programming Languages>Kendr CodeNew to Visual Studio Code? Get it now.
Kendr Code

Kendr Code

Kendr

|
6 installs
| (1) | Free
Standalone Kendr AI coding agents for VS Code with local and hosted models, rich chat, durable sessions, worktrees, previews, approvals, plans, and checkpoints.
Installation
Launch VS Code Quick Open (Ctrl+P), paste the following command, and press enter.
Copied to clipboard
More Info

Kendr Code for VS Code

Kendr Code brings a standalone agentic coding workspace into VS Code. Use local, self-hosted, or hosted models; keep multiple coding sessions; review changes; approve tools; work in isolated Git worktrees; and return to durable conversations without leaving the editor. Kendr Desktop is optional.

Highlights

  • Chat on the right. Open Kendr Code from its logo in VS Code's secondary sidebar and keep the conversation beside your editor.
  • Editor toolbar launcher. Open Kendr from its official logo at the top right of any file editor.
  • Sessions on the left. Use the Kendr Activity Bar logo to create, find, switch, rename, refresh, or expand durable coding sessions.
  • Multiple conversations. The right sidebar has a dedicated searchable Chats screen. Select a row to enter its thread and use Back to return to the list without replacing editor documents.
  • Cross-project session library. Search conversations across every Kendr project, keep important work pinned, archive completed work, or fork a settled conversation into another project.
  • Optional editor tabs. Open any conversation in a retained editor tab when you need more room or a side-by-side layout.
  • Automatic background runtime. Kendr first reuses a saved kendr serve connection or a running Kendr Desktop app, then silently falls back to its bundled authenticated runtime. There is no runtime chooser in the normal workflow.
  • Local and hosted model routes. Add Ollama, LM Studio, vLLM, llama.cpp, OpenAI-compatible endpoints, OpenAI, Anthropic, Google AI, xAI/Grok, or Kendr Hosted from VS Code, with Agent/Ask/Plan compatibility enforced in the composer.
  • Reviewable coding work. Follow streaming responses, tool activity, plans, todos, approvals, changed files, checkpoints, memories, previews, and delegated-agent progress in one conversation.
  • Native review. Open each changed file in VS Code's side-by-side diff editor, edit an agent's plan as Markdown, and explicitly start implementation from the reviewed plan.
  • Controlled context. Type @ to add workspace files, see usage against the selected model's context window, and compact a long conversation without starting over.
  • Resilient long-running work. Retry-safe requests use bounded recovery, while interrupted event streams resume from the last durable event without replaying completed activity.

Getting started

  1. Install the Kendr Code package for your operating system and CPU, then reload VS Code when prompted.
  2. Open a local project folder in VS Code. The bundled runtime starts automatically; Kendr Desktop is not required.
  3. Click the Kendr logo on the right to open the Chats screen, then choose a conversation or select New chat.
  4. Click the Kendr logo on the left to open Sessions, then start a new session or select an existing one.
  5. Review the displayed project path and explicitly trust it when you want Kendr to work with that folder.
  6. Select Add Model or Provider Account, connect a subscription, API, hosted service, or local runtime, and choose a model on one of its compatible interaction modes. Kendr account sign-in uses the browser with an email-code fallback.

Runtime discovery checks a saved Kendr server first, then a healthy Kendr Desktop session, and otherwise starts the bundled local runtime. Failed probes fall through automatically. External tokens are stored in VS Code SecretStorage, not in settings or workspace files. Connect to Kendr Server remains in the Command Palette for intentionally managed or remote kendr serve processes.

Sessions and chat

The left Sessions view is the durable conversation navigator. Its title actions create a session, search, and refresh the list. Conversations from every Kendr project are grouped into Pinned, Needs attention, Recent, Older, and Archived. Selecting a session focuses the right chat, switches to its project context, and attaches the chat to that conversation. Session context actions can pin, archive, unarchive, fork, rename, remove, or open a conversation in an editor tab. Forking can target another project without replacing the original conversation.

New Conversation opens an empty composer immediately and creates the durable session only when you send its first message. For Git projects, select Current workspace or Isolated worktree directly in that composer.

The right sidebar separates navigation from conversation work. Its Chats screen contains search and full conversation rows; selecting one opens a focused thread with safe rich Markdown, code and message copy actions, prompt history, slash commands, context chips, bounded file attachments, and an auto-sizing composer. Running turns can be stopped or steered without losing the session.

Type @ in the composer to search workspace paths and insert a file reference without leaving the keyboard. Kendr shows the current usage against the selected model's advertised context window and warns as it approaches the limit. Use /compact or Kendr Code: Compact Conversation Context to ask the backend to condense an inactive conversation while preserving its durable history.

Models and providers

The extension can create model connections through its own standalone backend or read the same safe catalog from Kendr Desktop. It supports local Ollama models and OpenAI-compatible local or self-hosted runtimes such as LM Studio, vLLM, and llama.cpp, as well as OpenAI, Anthropic, Google AI, xAI/Grok, Kendr Hosted, and supported subscription-runtime connections. Choose Provider subscription in chat or run Kendr Code: Connect Provider Subscription to see the reviewed account-login catalog. OpenAI Codex can use provider-managed ChatGPT login through the official runtime. Cataloged methods that are not safe or permitted to execute remain visible with their exact unavailable reason. The current OpenAI Codex preview appears in Ask and Plan in Kendr Code; it supports compatible research and code-review surfaces but does not enable full Agent mode or act as an OpenAI API-key replacement.

The model picker groups choices by connection and labels each route Local or Cloud. It shows only ready, enabled, policy-eligible routes compatible with the selected Agent, Ask, or Plan mode. Choose Find local models with Ollama directly in the picker for local discovery, or Connect another model provider for subscriptions, APIs, and other runtimes. The only + in the composer is the workspace-file attachment control.

Kendr Hosted exposes every entitled text-generation route to Agent mode. Models with native tool calling use it directly; other hosted text models use Kendr's constrained coding-action transport, so the model list is not artificially limited to tool-advertising routes.

Use Kendr Code: Detect Local Models to test the standard loopback endpoints for Ollama, LM Studio, vLLM, and llama.cpp and connect a detected server without typing its URL.

Provider credentials, endpoints, and billing identities remain backend-side; they are never copied into the chat webview or regular VS Code settings. A turn or inline completion carries only the selected model, connection, and execution-route identifiers, preventing a same-named model on another account from being selected accidentally.

Coding capabilities

  • Git-aware local and isolated-worktree sessions.
  • Explicit active-file, selection, diagnostics, rich @ file references, and bounded multi-file context.
  • Model-aware context-window usage, warning thresholds, and manual compaction.
  • Inline approvals and structured questions from the coding agent.
  • Native per-file diff review, editable plan review, checkpoints, and conversation rewind.
  • Pinned, archived, forked, and cross-project conversations.
  • Project branches, skills, MCP assignments, and capability controls.
  • Editable workspace memories and delegated-agent activity.
  • Detected app previews opened through VS Code's Simple Browser.
  • Optional fill-in-the-middle and same-line next-edit suggestions.

Commands

Command Action
Kendr Code: Open Chat Open or focus the right Kendr chat
Kendr Code: Open Conversation Select a durable session and open its chat
Kendr Code: Open Conversation in Editor Open the selected session in a retained editor tab
Kendr Code: New Conversation Create a local or isolated-worktree session
Kendr Code: Rename Conversation Rename a durable session
Kendr Code: Search Conversations Search and switch durable sessions
Kendr Code: Pin Conversation Keep a conversation at the top of Sessions
Kendr Code: Unpin Conversation Return a pinned conversation to its normal group
Kendr Code: Archive Conversation Move completed work into Archived
Kendr Code: Unarchive Conversation Restore an archived conversation
Kendr Code: Fork Conversation Continue a settled conversation in a selected project
Kendr Code: Remove Conversation Remove a session after explicit confirmation
Kendr Code: Show Workspace Changes Open the selected session's Git diff
Kendr Code: Open Plan for Review Open the current plan as an editable Markdown document
Kendr Code: Start Plan Implementation Save the reviewed plan and start its implementation
Kendr Code: Compact Conversation Context Condense an inactive conversation's working context
Kendr Code: Open Changes in Source Control Open VS Code's native Source Control view
Kendr Code: Open App Preview Open a detected application preview
Kendr Code: Trust Current Project Trust the displayed folder path explicitly
Kendr Code: Stop Active Turn Stop the current agent turn
Kendr Code: Toggle Inline Completions Enable or disable predictive ghost text
Kendr Code: Refresh Refresh projects, sessions, models, and conversation state
Kendr Code: Add Model or Provider Account Connect a subscription account, local runtime, hosted service, or API provider
Kendr Code: Connect Provider Subscription Open the provider-owned subscription login catalog
Kendr Code: Detect Local Models Find supported local model servers on standard endpoints
Kendr Code: Refresh Models Reload ready local and cloud execution routes
Kendr Code: Connect to Kendr Server Save the bearer token for an intentionally managed kendr serve process; startup still falls back automatically if it is unavailable
Kendr Code: Sign In to Kendr Sign in and enable entitled hosted models
Kendr Code: Sign Out of Kendr Revoke the VS Code account session

Security and privacy

Kendr Code does not infer project trust from an open folder. File changes, approvals, capability changes, and checkpoint restores remain explicit and are handled by the Kendr backend's audited authority paths. The standalone service binds only to a random loopback port, uses a new random bearer token for every launch, verifies the bundled executable against its packaged build manifest, and stops when its owning VS Code extension host exits.

Attachments are restricted to files inside the selected workspace, reject binary and unavailable files, and use bounded context budgets. Hosted-provider credentials and local-provider configuration remain backend-side. Account sessions and local API tokens are kept in VS Code SecretStorage and are not exposed to the chat webview.

Browser sign-in uses OAuth authorization code flow with PKCE. The callback returns only to the authenticated loopback runtime, and refresh tokens remain in VS Code SecretStorage.

Retry-safe API calls use bounded deadlines, server-directed backoff, and idempotency protection. Event streams detect stale connections and resume from the last durable cursor with duplicate suppression, so reconnecting does not repeat completed events.

Inline completion is disabled by default. When enabled, it sends only a bounded prefix and suffix from the visible document plus non-content request metadata.

Remote workspaces

For Remote SSH, dev containers, or WSL, install the matching Kendr Code platform package in the remote workspace extension host. Its bundled backend then runs in the same environment as the workspace. External runtime mode remains available when you intentionally operate a separate kendr serve process.

Compatibility

Kendr Code supports VS Code 1.90 and newer. Standalone release packages are produced for Windows x64 and arm64, Linux x64 and arm64, and macOS x64 and arm64. Release automation verifies platform signing or notarization where the operating system provides it, verifies signed checksums and provenance for Linux, and publishes checksums, a CycloneDX SBOM, and SLSA/in-toto provenance with each release set. VS Code 1.106 and newer show the dedicated right-sidebar chat launcher. Earlier supported versions keep Sessions and a Chat fallback in the left Activity Bar.

Support

Visit kendr.org or email support@kendr.org. The public extension is available on the Visual Studio Marketplace.

  • Contact us
  • Jobs
  • Privacy
  • Manage cookies
  • Terms of use
  • Trademarks
  • Your Privacy Choices
  • Consumer Health Privacy
© 2026 Microsoft