Commands Reference

All commands are available via the Command Palette (Ctrl+Shift+P). Type "MR Multi coder" to filter.

Chat & Navigation

Command	Keybinding	Description
`mrMultiCoder.openChat`	`Ctrl+Shift+A`	Open or reveal the chat panel in the sidebar.
`mrMultiCoder.focusInput`	`Ctrl+Escape`	Toggle focus between the editor and the chat input.
`mrMultiCoder.openChatInEditor`	`Ctrl+Shift+Alt+A`	Open a new chat tab in the editor area.
`mrMultiCoder.newSession`		Start a fresh session.
`mrMultiCoder.stopGeneration`	`Escape Escape`	Abort the current AI response (only while generating).
`mrMultiCoder.continueResponse`		Ask the model to continue from where it stopped.
`mrMultiCoder.openChatList`		Open the list of all conversations.
`mrMultiCoder.newChatFromEditor`		Start a new chat from the editor title bar.

Sessions

Command	Description
`mrMultiCoder.resumeSession`	Pick and resume a past conversation from the history list.
`mrMultiCoder.openSessionInTab`	Open a specific session in a new editor tab.
`mrMultiCoder.renameSession`	Rename the current conversation.
`mrMultiCoder.deleteSession`	Delete a session from history.
`mrMultiCoder.forkSessionWithSummary`	Fork the current session with an LLM-generated summary.
`mrMultiCoder.compactConversation`	Compact the session — summarises older turns to free context.
`mrMultiCoder.rewindToCheckpoint`	Rewind the conversation to a prior checkpoint.
`mrMultiCoder.openMemoryPanel`	Show the loaded project memory (CODER.md and friends).
`mrMultiCoder.showSessionMemory`	Open this session's MEMORY.md with compaction diagnostics.

Provider & Model Selection

Command	Description
`mrMultiCoder.selectProvider`	Switch the active AI provider (Claude, MR Coder, Kilo Code, Qwen, OpenRouter, Ollama, Codex, Gemini, HuggingFace).
`mrMultiCoder.selectModel`	Switch the model within the current provider.
`mrMultiCoder.selectPermissionMode`	Change permission mode: default, plan, acceptEdits, auto, or bypassPermissions. `auto` auto-approves tools but AskUserQuestion still waits for the user; `bypassPermissions` auto-approves tools AND auto-answers questions with the recommended option.

Authentication

Command	Description
`mrMultiCoder.setClaudeApiKey`	Store your Anthropic API key in secure storage.
`mrMultiCoder.clearClaudeApiKey`	Remove the stored Claude API key.
`mrMultiCoder.validateClaudeApiKey`	Test the stored Claude key with a lightweight API call.
`mrMultiCoder.loginWithClaude`	Sign in with your claude.ai account (OAuth) instead of an API key.
`mrMultiCoder.switchClaudeAuth`	Switch between claude.ai OAuth and a stored API key.
`mrMultiCoder.connectMrcoderOAuth`	Connect your MR Coder account via OAuth — provisions a key automatically.
`mrMultiCoder.setMrcoderApiKey`	Store an existing MR Coder API key (sk-mr-auth-…).
`mrMultiCoder.clearMrcoderApiKey`	Remove the stored MR Coder API key.
`mrMultiCoder.loginQoder`	Sign in with Qoder — reads the CLI's token from your OS keychain.
`mrMultiCoder.setQwenApiKey`	Store your DashScope (Qwen) API key.
`mrMultiCoder.clearQwenApiKey`	Remove the stored Qwen API key.
`mrMultiCoder.validateQwenApiKey`	Test the stored Qwen key against DashScope.
`mrMultiCoder.setOpenrouterApiKey`	Store your OpenRouter API key.
`mrMultiCoder.clearOpenrouterApiKey`	Remove the stored OpenRouter API key.
`mrMultiCoder.validateOpenrouterApiKey`	Test the stored OpenRouter key — hits /auth/key, costs zero tokens.
`mrMultiCoder.setCodexApiKey`	Store your OpenAI API key.
`mrMultiCoder.clearCodexApiKey`	Remove the stored OpenAI Codex API key.
`mrMultiCoder.validateCodexApiKey`	Test the stored OpenAI key.
`mrMultiCoder.loginWithCodex`	Sign in with the Codex CLI (OAuth) — auto-switches auth mode to CLI.
`mrMultiCoder.setGeminiApiKey`	Store your Google AI Studio API key.
`mrMultiCoder.clearGeminiApiKey`	Remove the stored Gemini API key.
`mrMultiCoder.validateGeminiApiKey`	Test the stored Gemini key.
`mrMultiCoder.loginWithGemini`	Sign in with the Gemini CLI using your Google Account.
`mrMultiCoder.setHuggingfaceApiKey`	Store your HuggingFace API token.
`mrMultiCoder.clearHuggingfaceApiKey`	Remove the stored HuggingFace token.
`mrMultiCoder.validateHuggingfaceApiKey`	Test the stored HuggingFace token.
`mrMultiCoder.authenticateCliProvider`	Sign in to a CLI-based provider (Kilo Code, Qwen, Codex, Gemini).
`mrMultiCoder.connectAccount`	Connect to mr-coder.io.

API keys are stored securely in your OS keychain — never in settings.json, logs, or synced storage.

Code Context

Command	Keybinding	Description
`mrMultiCoder.addSelectionToChat`	`Ctrl+Shift+L`	Attach the current editor selection to the chat as context.
`mrMultiCoder.addFileToChat`		Attach the current file to the chat (from explorer right-click).
`mrMultiCoder.insertMentionAtSelection`	`Alt+K`	Insert an @file mention for the current selection into the chat input.
`mrMultiCoder.explainSelection`		Attach selection and ask the model to explain it.
`mrMultiCoder.refactorSelection`		Attach selection and ask the model to refactor it.

Analytics

Command	Description
`mrMultiCoder.OpenAnalytics`	Open the Developer Analytics Dashboard.

Full analytics documentation in the Analytics Dashboard page.

Web Remote Access & Tunnel

Command	Description
`mrMultiCoder.serveWeb`	Start the VS Code Web server so you can reach this workspace from a browser.
`mrMultiCoder.stopServeWeb`	Stop the VS Code Web server.
`mrMultiCoder.addWebUser`	Create a web-access user (password is stored only as a salted hash).
`mrMultiCoder.manageWebUsers`	List, edit, or remove web-access users and their project permissions.
`mrMultiCoder.tunnel.create`	Provision a public tunnel domain for the auth proxy.
`mrMultiCoder.tunnel.status`	Report whether the tunnel is live.
`mrMultiCoder.tunnel.disconnect`	Tear the tunnel down.

Full setup and the security model in the Web Remote Access guide.

Product Owner Agent

Command	Description
`mrMultiCoder.po.captureSnapshot`	Capture a PO snapshot of the current turn manually.
`mrMultiCoder.po.openSnapshot`	Open a saved PO snapshot.
`mrMultiCoder.po.shareSnapshot`	Share a PO snapshot to your configured endpoint.
`mrMultiCoder.po.revealSnapshot`	Reveal a PO snapshot in your OS file manager.
`mrMultiCoder.po.deleteSnapshot`	Delete a PO snapshot.
`mrMultiCoder.po.refreshSnapshots`	Refresh the PO Snapshots sidebar view.

What the Product Owner does, and how to enable it, on the Agents page.

MCP (Model Context Protocol)

Command	Description
`mrMultiCoder.manageMcpServers`	Open the MCP server configuration panel.
`mrMultiCoder.toggleMcpServer`	Enable or disable a specific MCP server.

Diagnostics

Command	Description
`mrMultiCoder.showOutputChannel`	Open the "MR Multi coder" output channel.
`mrMultiCoder.openErrorLog`	Open the extension error log.
`mrMultiCoder.openSessionStorage`	Open the session storage folder in the file explorer.
`mrMultiCoder.openSettings`	Open the MR Multi coder settings panel.
`mrMultiCoder.openSpeechSettings`	Open Settings at Speech to Text & Text to Speech — your Deepgram API key, voice and auto-play.
`mrMultiCoder.stopReadingAloud`	Stop the reply being read aloud.
`mrMultiCoder.toggleVoiceInput`	Push-to-talk: start recording in the focused chat, or stop and insert the text (Ctrl+Alt+V / ⌘⌥V).
`mrMultiCoder.stopListening`	Stop listening for the wake word in this window (the status-bar item does the same).
`mrMultiCoder.toggleMicMute`	Mute or unmute the microphone. While muted nothing is recorded — not push-to-talk, not the wake word — until you unmute.
`mrMultiCoder.runVerification`	Run typecheck, tests, and lint over the workspace.
`mrMultiCoder.runUxScreenshots`	Capture UX screenshots for an end-user UI review.
`mrMultiCoder.rollout.status`	Report rollout kill-switch status (debug).

Haptic Feedback & Sounds

MR Multi coder plays audio cues during the chat lifecycle. All sounds work out of the box with built-in synthesized tones. You can override any sound with your own audio file.

Sound Events

Event	Sound	When
Submit	Short click	When you press Enter to send
Response Complete	Single ding	Model finishes a response (no tool use)
Task Complete	Double chime	Model finishes with tool calls
Question	Double tone	Model asks for approval
Error	Low thud	Provider encounters an error
Progress	Ambient tone	Plays continuously while generating

Haptic Commands

Command	Description
`mrMultiCoder.configureHapticSound`	Pick a sound event and assign a custom audio file (.wav, .mp3, .ogg, .m4a).
`mrMultiCoder.configureProgressFolder`	Pick a folder of audio files for progress music during generation.
`mrMultiCoder.resetHapticFeedback`	Reset all haptic settings to defaults.

Progress Music

Play background music while the AI is generating. Requires both the master switch and the progress switch enabled.

1. Create a folder with your audio files (.wav, .mp3, .ogg, .m4a)

2. Run Configure Progress Music Folder and select it

3. Enable mrMultiCoder.hapticFeedback.onProgress.enabled

Multiple tabs: only one tab plays at a time. Music hands off automatically when tabs finish.

Slash Commands (In-Chat)

Type these directly in the chat input — no Command Palette needed.

Command	Description
`/compact`	Summarise earlier turns to free context window space.
`/clear`	Start a new empty session (old one stays in history).
`/new`	Alias for /clear.
`/fork <summary>`	Fork into a new session seeded with a summary.
`/init`	Interactive wizard to set up project conventions (CODER.md).
`/cancel`	Cancel a running in-chat wizard.
`/help`	Show the list of available slash commands.
`/cost`	Show current session's token usage and estimated cost.
`/schedule`	Schedule a message to send later.
`/schedule list`	View pending scheduled messages.
`/schedule cancel <id>`	Cancel a pending schedule.
`/mcp`	Connect an MCP server — Jira, Figma, Stitch, or any HTTP endpoint.
`/memory`	Inspect loaded project memory. Also: /memory open \| global \| reload.
`/po`	Product Owner agent: enable \| disable \| status \| mode \| review \| risks \| next.
`/agents`	Open the Agents panel (Product Owner and more).
`/agent`	Alias for /agents.
`/usage`	Show MR Coder token usage (rolling window, weekly, monthly).
`/autospeak on \| off \| status`	Read each reply of this chat aloud when it finishes (per chat — other chats are not affected). Uses Deepgram voices with your key (Settings → Speech), or the free system voice.
`/hello-coder on \| off \| status`	Hands-free wake word: say “hello coder, <request>” and it is sent. Uses your Deepgram key (Settings → Speech). Listening never starts by itself.