Kilo Code CLI
Orchestrate agents from your terminal. Plan, debug, and code fast with keyboard-first navigation on the command line.
The Kilo Code CLI uses the same underlying technology that powers the IDE extensions, so you can expect the same workflow to handle agentic coding tasks from start to finish.
Install
npm install -g @kilocode/cli
Change directory to where you want to work and run kilocode:
# Start interactive chat session
kilocode
# Start with a specific mode
kilocode --mode architect
# Start with a specific workspace
kilocode --workspace /path/to/project
to start the CLI and begin a new task with your preferred model and relevant mode.
What you can do with Kilo Code CLI
- Plan and execute code changes without leaving your terminal. Use your command line to make edits to your project without opening your IDE.
- Switch between hundreds of LLMs without constraints. Other CLI tools only work with one model or curate opinionated lists. With Kilo, you can switch models without booting up another tool.
- Choose the right mode for the task in your workflow. Select between Architect, Ask, Debug, Orchestrator, or custom agent modes.
- Automate tasks. Get AI assistance writing shell scripts for tasks like renaming all of the files in a folder or transforming sizes for a set of images.
CLI reference
CLI commands
Command | Description | Example |
---|---|---|
kilocode | Start interactive | |
/mode | Switch between modes (architect, code, debug, ask, orchestrator) | /mode orchestrator |
/model | Learn about available models and switch between them | |
/model list | List available models | |
/model info | Prints description for a specific model by name | /model info z-ai/glm-4.5v |
/model select | Select and switch to a new model | |
/new | Start a new task with the agent with a clean slate | |
/help | List available commands and how to use them | |
/exit | Exit the CLI |
Config reference for providers
Kilo gives you the ability to bring your own keys for a number of model providers and AI gateways, like OpenRouter and Vercel AI Gateway. Each provider has unique configuration options and some let you set environment variables.
You can reference the Provider Configuration Guide for examples if you want to edit .config files manually. You can also run:
kilocode config
to complete configuration with an interactive workflow on the command line.
Autonomous mode (Non-Interactive)
Autonomous mode allows Kilo Code to run in automated environments like CI/CD pipelines without requiring user interaction.
# Run in autonomous mode with a prompt
kilocode --auto "Implement feature X"
# Run in autonomous mode with piped input
echo "Fix the bug in app.ts" | kilocode --auto
# Run in autonomous mode with timeout (in seconds)
kilocode --auto "Run tests" --timeout 300
Autonomous Mode Behavior
When running in Autonomous mode (--auto
flag):
- No User Interaction: All approval requests are handled automatically based on configuration
- Auto-Approval/Rejection: Operations are approved or rejected based on your auto-approval settings
- Follow-up Questions: Automatically responded with a message instructing the AI to make autonomous decisions
- Automatic Exit: The CLI exits automatically when the task completes or times out
Auto-Approval Configuration
Autonomous mode respects your auto-approval configuration. Edit your config file with kilocode config
to customize:
{
"autoApproval": {
"enabled": true,
"read": {
"enabled": true,
"outside": true
},
"write": {
"enabled": true,
"outside": false,
"protected": false
},
"execute": {
"enabled": true,
"allowed": ["npm", "git", "pnpm"],
"denied": ["rm -rf", "sudo"]
},
"browser": {
"enabled": false
},
"mcp": {
"enabled": true
},
"mode": {
"enabled": true
},
"subtasks": {
"enabled": true
},
"question": {
"enabled": false,
"timeout": 60
},
"retry": {
"enabled": true,
"delay": 10
},
"todo": {
"enabled": true
}
}
}
Configuration Options:
read
: Auto-approve file read operationsoutside
: Allow reading files outside workspace
write
: Auto-approve file write operationsoutside
: Allow writing files outside workspaceprotected
: Allow writing to protected files (e.g., package.json)
execute
: Auto-approve command executionallowed
: List of allowed command patterns (e.g., ["npm", "git"])denied
: List of denied command patterns (takes precedence)
browser
: Auto-approve browser operationsmcp
: Auto-approve MCP tool usagemode
: Auto-approve mode switchingsubtasks
: Auto-approve subtask creationquestion
: Auto-approve follow-up questionsretry
: Auto-approve API retry requeststodo
: Auto-approve todo list updates
Command Approval Patterns
The execute.allowed
and execute.denied
lists support hierarchical pattern matching:
- Base command:
"git"
matches any git command (e.g.,git status
,git commit
,git push
) - Command + subcommand:
"git status"
matches any git status command (e.g.,git status --short
,git status -v
) - Full command:
"git status --short"
only matches exactlygit status --short
Example:
{
"execute": {
"enabled": true,
"allowed": [
"npm", // Allows all npm commands
"git status", // Allows all git status commands
"ls -la" // Only allows exactly "ls -la"
],
"denied": [
"git push --force" // Denies this specific command even if "git" is allowed
]
}
}
Interactive Command Approval
When running in interactive mode, command approval requests now show hierarchical options:
[!] Action Required:
> ✓ Run Command (y)
✓ Always run git (1)
✓ Always run git status (2)
✓ Always run git status --short --branch (3)
✗ Reject (n)
Selecting an "Always run" option will:
- Approve and execute the current command
- Add the pattern to your
execute.allowed
list in the config - Auto-approve matching commands in the future
This allows you to progressively build your auto-approval rules without manually editing the config file.
Autonomous Mode Follow-up Questions
In Autonomous mode, when the AI asks a follow-up question, it receives this response:
"This process is running in non-interactive Autonomous mode. The user cannot make decisions, so you should make the decision autonomously."
This instructs the AI to proceed without user input.
Exit Codes
0
: Success (task completed)124
: Timeout (task exceeded time limit)1
: Error (initialization or execution failure)
Example CI/CD Integration
# GitHub Actions example
- name: Run Kilo Code
run: |
echo "Implement the new feature" | kilocode --auto --timeout 600
Local Development
DevTools
In order to run the CLI with devtools, add DEV=true
to your pnpm start
command, and then run npx react-devtools
to show the devtools inspector.