Qodo CLI
Installation
# install with npm
npm install -g @qodo/command
Usage
Usage
$ qodo [options] <prompt>
$ qodo run <command> [extra instructions]
Commands
run <command> Execute a specific agent command with optional extra instructions
login Log in to Qodo
models Get available models
chat Start an interactive chat session
key list List all API keys
key create <name> Create a new API key with the given name
key revoke <name> Revoke an API key by name
create-agent Create a new agent by translating user requirements into a valid configuration
list-agents List available agents from configuration (interactive in CLI mode)
list-mcp List available local and remote tools
self-review Analyze git changes and group them into logical change groups (opens web interface)
update Manage auto-updates (--check, --install, --configure)
-h, --help Show help and exit
chain "A > B > C" Run multiple agents sequentially (quote the chain!)
-v, --version Show version and exit
-l, --log=path Redirect console output to a file | stdout | stderr
-y, --yes Confirm all prompts automatically (useful for CI)
-q, --silent Suppress all console output except the final result (logs go to /dev/null)
-d, --debug Enable debug mode with verbose logging and no data truncation
--dir=path Specify project root directories (can be used multiple times)
--ci Run commands in CI mode
--mcp Run commands as tools from agent config in MCP-server-like mode
--ui Open Qodo with web interface
--webhook Run commands as tools from agent config in webhook mode
--slack Run as Slack bot (HTTP webhook mode by default)
-p, --port=number Specify custom port for server modes (webhook, slack, mcp, ui)
--plan Use planning execution strategy (agent plans before acting)
--act Use direct execution strategy (agent acts immediately)
-m, --model=model_name Specify a custom model to use
--agent-file=path Specify a custom path to agent configuration file
--mcp-file=path Specify a custom path to mcp.json
-r, --resume=session_id Resume a task with the given session ID
--set key=value Set custom key-value pairs (can be used multiple times)
--no-builtin Disable built-in MCP servers (ripgrep, filesystem, git)
-t, --tools=list Specify authorized tools (comma-separated, e.g., shell,git,filesystem)
Note: Some tools have dependencies (e.g., qodo_merge requires git)
--tool <name> Specify authorized tool (can be used multiple times, e.g., --tool shell --tool git --tool filesystem)
--permissions=level Set permissions level (r=read, rw=read+write, rwx=full, -=none)
--with=session_id Preload context with a previous session summarization (used before task/hotstart)
--sandbox=mode Run shell commands in sandbox (macOS only) [off|permissive|restrictive|path/to/profile.sb] (default: off)
Examples
$ qodo chain "improve > review > open-pr"
$ qodo chain "improve > review > open-pr" --chain-step-timeout=600000
$ qodo chain "improve > review > open-pr" --chain-continue-on-error
$ qodo "Review my latest changes and suggest improvements"
$ qodo run review
$ qodo run review "focus on security issues"
$ qodo review --set coverage_score_threshold=0.8
$ qodo self-review
$ qodo "Analyze code" --dir /path/to/project1 --dir /path/to/project2 --permissions=r
$ qodo chat
$ qodo key list
$ qodo key create my-ci-key
$ qodo key revoke my-old-key
$ qodo create-agent --set goal="review code" --set description="analyze pull requests"
$ qodo list-agents
$ qodo list-mcp
$ qodo --ui
$ qodo --tools=shell,filesystem "analyze this codebase"
$ qodo --sandbox=./custom-profile.sb "run a shell command with custom sandbox"