Kimi Code PLACEMAT

Keyboard & TUI

General input

Item	Description
#Enter	Submit the current input
#Shift-Enter / Ctrl-J	Insert a newline in the input
#Up / Down	Browse input history
#Esc	Close a popup, cancel completion, or interrupt streaming and compaction
#Ctrl-C	Interrupt streaming output, or clear the input box +3
#Ctrl-D	Exit when the input box is empty +1
#Ctrl-T	Expand or collapse the todo list when it is truncated
#Ctrl-P / Ctrl-N	Page through the experimental Updates panel

Mode switching

Item	Description
#Shift-Tab	Toggle Plan mode +3
#!	Enter shell mode from an empty input box +1

Input & editing

Item	Description
#Ctrl-G	Edit the current input in an external editor +2
#Ctrl-V	Paste an image or video from the clipboard (Unix, macOS) +2
#Ctrl--	Undo
#Esc Esc	Open the undo selector (double-press while idle)

During streaming

Item	Description
#Ctrl-S	Steer: inject the current input into the running turn +3

Tool output

Item	Description
#Ctrl-O	Expand or collapse tool output, shell output and compaction summaries +1

Approval panel

Item	Description
#1 / 9	Select the option at that index directly
#Ctrl-E	Expand or collapse a diff or file preview in the panel
#Esc / Ctrl-C / Ctrl-D	Reject the current request +1

Slash Commands (Core)

Account & configuration

Item	Description
#/login	Pick an account or platform and sign in +1
#/logout	Clear credentials for the selected account
#/provider	Open the interactive provider manager
#/model	Switch the model used in this session
#/secondary-model / /subagent-model	Pick the default model for subagents +1
#/settings / /config	Open the settings panel in the TUI
#/permission	Select a permission mode
#/editor	Configure the external editor launched by Ctrl-G
#/theme	Switch the terminal colour theme
#/experiments / /experimental	Open the experimental feature panel

Session management

Item	Description
#/new / /clear	Start a fresh session, discarding the current context
#/sessions / /resume	Browse and switch to a historical session +1
#/fork	Fork a new session from this one, keeping full history +1
#/title / /rename	Show or set the session title (max 200 characters)
#/compact	Compact the conversation context to free tokens +1
#/undo	Undo recent prompts from the active context +2
#/reload / /reload-tui	Reload the session and apply the latest config.toml and tui.toml +1
#/tasks / /task	Browse the background task list
#/add-dir	Add an extra workspace directory to this session +2
#/exit / /quit / /q	Exit the CLI

Modes & run control

Item	Description
#/yolo / /yes	Preselect Ask When Needed in the permission list +3
#/auto	Preselect Never Ask in the permission list +1
#/plan	Toggle Plan mode, or force it with on / off +1

Export & sharing

Item	Description
#/export-md / /export	Export the session as a Markdown file
#/export-debug-zip	Export the session as a debug ZIP archive
#/copy	Copy the last assistant message to the clipboard +1
#/web	Open the current session in the web UI +1
#/desktop / /install-desktop	Open the Kimi Code desktop app page in your browser

Information & status

Item	Description
#/help / /h / /?	Show keyboard shortcuts and all available commands +1
#/status	Show version, model, working directory and permission mode
#/usage	Show token usage, context consumption and quota
#/version	Display the CLI version number
#/btw	Open a side conversation in a forked sub-agent +1
#/feedback / /bug	Submit feedback with optional diagnostic logs and codebase context

Slash Commands (Tools)

Swarm & goals

Item	Description
#/swarm on / /swarm off	Turn swarm mode on or off without sending a prompt
#/swarm <task>	Turn swarm mode on and send the task as a prompt +2
#/goal	Start or manage an autonomous goal +3
#/goal pause / /goal resume / /goal cancel / /goal replace	Pause, resume, cancel or replace the saved goal
#/goal next	Queue an upcoming goal for this session +3

Extensibility

Item	Description
#/mcp	List MCP servers and their connection status
#/plugins	Open the interactive plugin manager
#/init	Analyse the codebase and generate AGENTS.md

Built-in skill commands

Item	Description
#/mcp-config	Configure MCP servers and handle MCP OAuth login
#/custom-theme	Create or edit a custom TUI colour theme
#/update-config	Inspect or edit config.toml and tui.toml
#/check-kimi-code-docs	Answer Kimi Code product questions against the official docs
#/import-from-cc-codex	Import Claude Code and Codex instructions, skills and MCP settings
#/sub-skill	Reorganise the local skill inventory into hierarchical bundles +1

Skill commands

Item	Description
#/skill:<name>	Invoke an activated external skill +3
#/<parent>.<sub-skill>	Invoke an external sub-skill by its dotted name

CLI & Subcommands

Flags

Item	Description
#-V, --version / -h, --help	Print the version number
#-S, --session [id]	Resume a session; with no id, pick interactively
#-c, --continue	Continue the previous session for this working directory
#-m, --model <model>	Model alias for this invocation +1
#-p, --prompt <prompt>	Run one prompt non-interactively and print the response +2
#--output-format <format>	Output format for prompt mode: text or stream-json
#-y, --yolo	Start in Ask When Needed mode +2
#--auto	Start in Never Ask mode — nothing interrupts you
#--plan	Start in Plan mode
#--skills-dir <dir>	Load skills from this directory instead of auto-discovery; repeatable
#--agent <name>	Agent profile for the new session +1
#--agent-file <path>	Load an agent definition from a Markdown file for this launch +1
#--add-dir <dir>	Add an extra workspace directory for this session; repeatable

Subcommands

Item	Description
#kimi export	Export a session as a ZIP archive
#kimi fork	Fork a session into a new session
#kimi session	Manage sessions non-interactively
#kimi provider	Manage LLM providers non-interactively
#kimi login	Authenticate via the device-code flow
#kimi doctor	Validate the configuration files
#kimi web	Run the local Kimi server and open the web UI +1
#kimi server	Deprecated — use kimi web instead
#kimi acp	Run Kimi Code as an Agent Client Protocol server over stdio
#kimi vis	Launch the session visualiser in your browser
#kimi install-desktop	Print the desktop app page and open it in your browser +1
#kimi migrate	Migrate data from a legacy kimi-cli installation
#kimi upgrade / kimi update	Upgrade Kimi Code to the latest version +1

Settings (config.toml)

Files

Item	Description
#~/.kimi-code/config.toml	Main config, created on first run +2
#~/.kimi-code/tui.toml	Terminal-UI and client preferences, alongside config.toml +2
#.kimi-code/local.toml	Project-local settings that should not be shared +2
#$KIMI_CODE_HOME/SYSTEM.md	Permanently replaces the main agent’s system prompt +2
#kimi doctor	Validate the configuration files

Top-level

Item	Description
#default_model	Default model alias; must be defined in models
#default_permission_mode	Permission mode for new sessions: manual, yolo or auto
#default_plan_mode	Whether new sessions start in Plan mode
#merge_all_available_skills	Merge skills from all available directories
#extra_skill_dirs / extra_agent_dirs	Extra skill search directories layered on the defaults
#builtin_product_skills	Offer the built-in skills documenting Kimi Code to the model
#telemetry	Anonymous telemetry; disabled only when explicitly set to false

Models & providers

Item	Description
#[providers."<id>"]	API provider entry: type, base_url, api_key +1
#[models."<alias>"]	Model alias entry +1
#[secondary_model]	Subagent model pool +2
#[thinking]	Thinking-mode defaults: enabled, effort, keep
#[token_counting]	How context tokens are reported externally

Runtime

Item	Description
#[loop_control]	Agent loop limits +2
#[background]	Background task runtime +1
#[subagent] / [swarm]	Subagent and AgentSwarm limits such as timeout_ms
#[tools]	Global tool switch
#[read]	Character limits and resumable long-line file reads +1
#[image]	Image compression parameters such as max_edge_px
#[database]	Session-index read model and global search worker +2
#[watch]	Filesystem watchers that reload local.toml, AGENTS.md, skills and MCP config +2
#[services]	Built-in external services +1
#[identity]	Custom agent identity — the name it calls itself and its client slug

Permissions & hooks

Item	Description
#[[permission.rules]]	Initial permission rules, each a decision plus a pattern +1
#[[hooks]]	Lifecycle hooks, declared as an array of tables

tui.toml

Item	Description
#theme	Colour theme: auto, dark, light or a custom theme name
#render_latex / disable_paste_burst	Render LaTeX in messages as Unicode text; false keeps the raw source +1
#cache_expiry_hint / disable_feedback_survey	Warn that the context cache may have expired on resume or after a long idle +2
#[editor].command	External editor command; empty falls back to $VISUAL / $EDITOR
#[notifications].enabled	Send desktop notifications +1
#[upgrade].auto_install	Install new versions automatically
#[status_line].items	Footer slots and order: mode, goal, model, tasks, cwd, git, tips +1
#[status_line].command	Custom status-line command; its first stdout line replaces the footer +1
#[markdown].mermaid	Render fenced mermaid code blocks in the terminal as diagrams; "off" disables it

Project-local

Item	Description
#[workspace].additional_dir	Extra absolute workspace directories for every session of this project

Environment Variables

Core paths

Item	Description
#KIMI_CODE_HOME	Data root; defaults to ~/.kimi-code +3
#KIMI_DISABLE_TELEMETRY	Turn off anonymous telemetry reporting
#KIMI_CODE_CUSTOM_HEADERS	Attach newline-separated Name: Value headers to every model request +2
#KIMI_CODE_PASSWORD	Parallel auth credential for kimi web +1

Models

Item	Description
#KIMI_MODEL_NAME	Synthesise a temporary in-memory provider and model, without touching config.toml +1
#KIMI_MODEL_THINKING_EFFORT	Force a thinking effort, bypassing the model’s declared support_efforts +1
#KIMI_MODEL_TEMPERATURE / KIMI_MODEL_TOP_P / KIMI_MODEL_MAX_COMPLETION_TOKENS	Sampling controls applied to every request +2

Runtime switches

Item	Description
#KIMI_CODE_BACKGROUND_KEEP_ALIVE_ON_EXIT	Keep background tasks when the session closes
#KIMI_CODE_BACKGROUND_MAX_RUNNING_TASKS	Cap concurrently running background tasks
#KIMI_CODE_AGENT_SWARM_MAX_CONCURRENCY	Cap AgentSwarm subagents during the initial ramp
#KIMI_SUBAGENT_TIMEOUT_MS / KIMI_CODE_SWARM_TIMEOUT_MS	Wall-clock ceiling for one Agent subagent
#KIMI_LOOP_MAX_STEPS_PER_TURN / KIMI_LOOP_MAX_ATTEMPTS_PER_STEP	Max agent steps per turn; 0 is unlimited
#KIMI_MCP_STARTUP_TIMEOUT_MS / KIMI_MCP_TOOL_TIMEOUT_MS	Default MCP connection timeout +1
#KIMI_CODE_PLUGIN_MARKETPLACE_URL	Override the marketplace JSON loaded by /plugins +1
#KIMI_CODE_NO_AUTO_UPDATE	Disable the update preflight entirely +1
#KIMI_DISABLE_CRON	Disable the scheduled-task tool +1
#KIMI_CODE_INFINITE_RETRY	Retry failed LLM requests indefinitely +1
#KIMI_CODE_PERMISSION_MODE_REMINDER	Set to 0 to stop injecting the auto permission-mode reminders into the model context
#KIMI_CODE_WATCH	Disable filesystem watchers for config and skill files +1

Experimental (churns between releases)

Item	Description
#KIMI_CODE_EXPERIMENTAL_FLAG	Enable every registered experimental feature for this process +1
#KIMI_CODE_EXPERIMENTAL_SUBAGENT_FORK	Start a subagent from a snapshot of the caller’s history
#KIMI_CODE_TUI_FULL_SCREEN	Experimental fullscreen UI +1
#KIMI_CODE_SEARCH_WORKER / KIMI_CODE_PERSISTENCE_MINIDB_READMODEL	Run the global search index in a dedicated worker thread +1

Diagnostics

Item	Description
#KIMI_LOG_LEVEL	Log level: off, error, warn, info, debug +2

Skills, Agents & AGENTS.md

Skills

Item	Description
#SKILL.md	The main file of a directory-form skill +3
#$KIMI_CODE_HOME/skills/	Kimi-specific user skills; moves with the data root
#~/.agents/skills/	Generic user skills, shared across tools +1
#.kimi-code/skills/ / .agents/skills/	Project skills
#--skills-dir	Load skills from an explicit directory instead of auto-discovery
#extra_skill_dirs	Extra skill search directories layered on the defaults +1

Agents

Item	Description
#$KIMI_CODE_HOME/agents/ / ~/.agents/agents/	Kimi-specific user agents
#.kimi-code/agents/ / .agents/agents/	Project agents
#override: true	Lets an agent file replace a same-name built-in agent +3
#${base_prompt}	Embeds the effective default system prompt inside an agent file +2
#--agent / --agent-file	Select an agent profile for a new session

AGENTS.md

Item	Description
#/init	Analyse the codebase and generate AGENTS.md
#AGENTS.md	Project instructions at the project root
#.kimi-code/AGENTS.md	Project-level Kimi-specific instructions
#$KIMI_CODE_HOME/AGENTS.md	Global Kimi-specific instructions; moves with the data root
#~/.agents/AGENTS.md	Generic cross-tool instructions in the real OS home

Plugins

Item	Description
#/plugins	Plugin manager with Installed, Official, Curated and Custom tabs +1
#KIMI_CODE_PLUGIN_MARKETPLACE_URL	Point /plugins at a different marketplace JSON

Hooks & MCP

Hook configuration

Item	Description
#[[hooks]]	One hook entry in config.toml +1
#matcher	Regex filtering which event targets fire the hook +1
#exit 0 / exit 2	Hook exit codes: 0 allows, 2 blocks +2

Hook events

Item	Description
#UserPromptSubmit / UserPromptQueued	Fires when you send a message; can block the model call this turn
#PreToolUse	Fires before a tool call, ahead of permission checks; blocking stops the tool
#PostToolUse / PostToolUseFailure	Fires after a tool succeeds
#PermissionRequest / PermissionResult	Fires just before waiting for approval
#SessionStart / SessionEnd	Fires after a session starts or resumes +1
#SessionHeartbeat	Fires every 60 seconds while the session is alive +1
#TurnStarted	Fires when a new turn begins, with its origin kind
#Stop / StopFailure	Fires when the model is about to end the turn; blocking lets it continue
#Interrupt	Fires when you interrupt the turn, in place of Stop +1
#SubagentStart / SubagentStop	Fires around sub-agent runs
#TaskStarted	Fires when a background task starts
#PreCompact	Fires before context compaction; return values are ignored entirely

MCP

Item	Description
#mcp.json	MCP server configuration, at user and project level +2
#mcp__<server>__<tool>	How MCP tools are named for permission rules +3
#[mcp] startup_timeout_ms / [mcp] tool_timeout_ms	Global MCP connection timeout, default 30000 ms +1
#kimi acp	Run Kimi Code as an Agent Client Protocol server over stdio
#kimi web	Run the local server and web UI +1