Usage

← Back to README

---

CLI Flags

Flag	Description
-m, --model <name>	Set the model
--provider <name>	Choose provider: ollama, openai, anthropic, qwen
-u, --ollama-url <url>	Ollama server URL
-p, --prompt <text>	Run a single prompt and exit
-r, --resume <id>	Resume a previous session
-c, --continue	Resume most recent session for this directory
-t, --team <preset>	Run with a multi-agent team (review, fullstack, security)
-e, --effort <level>	Set effort level: low, medium, high, max
-i, --image <path>	Attach an image to the prompt (base64-encoded)
--fast	Alias for --effort low
--browser	Enable Playwright browser tools
--no-buddy	Disable the buddy companion
--cwd <path>	Set working directory
--verbose	Print full tool output
--dangerously-skip-permissions	Auto-approve all tool calls
--max-turns <n>	Limit agent tool-call turns
-h, --help	Show help
-v, --version	Show version

Subcommands

cmdr serve

Start an HTTP API server.

cmdr serve --port 3120 --host 127.0.0.1 --model qwen3-coder:latest

Flag	Default	Description
--port <n>	3120	Server port
--host <addr>	127.0.0.1	Bind address
-m, --model <name>	—	Model to use

Endpoints:

Method	Path	Description
POST	/v1/chat	Send a message, get a JSON response
POST	/v1/stream	Send a message, get SSE stream
GET	/health	Health check (status, model, uptime)
GET	/v1/models	List available models

cmdr daemon

Background file watcher with command execution.

cmdr daemon start --watch src --on-change "npm test"
cmdr daemon status
cmdr daemon stop

Slash Commands

Command	Description
/help	Show available commands
/model <name>	Switch model
/models	List available Ollama models
/status	Show session info
/context	Show context window usage
/compact	Manually trigger compaction
/cost	Show token usage breakdown
/undo	Revert the last file change made by the agent
/diff	Show git diff
/review [staged|range|path]	AI-powered code review
/effort <level>	Set effort level (low, medium, high, max)
/fast	Alias for /effort low
/image <path>	Attach an image to the next prompt
/checkpoint save|list|restore|delete	Manage conversation checkpoints
/fork [name]	Fork the conversation into a branch
/branches	List conversation branches
/switch <branch>	Switch to a branch
/merge <branch>	Merge a branch into current
/index [path]	Index a directory for RAG search
/search <query>	Semantic search over indexed documents
/team [preset]	Switch to a multi-agent team
/agents	Show active agents and status
/tasks	Show task queue status
/config	View configuration
/plugin list	List loaded plugins
/mcp list	List MCP server connections
/session save	Save current session
/session resume <id>	Resume a session
/sessions	List saved sessions
/permissions [mode]	View/set permission mode
/init	Create CMDR.md template
/clear	Clear conversation
/quit	Exit

Built-in Tools

Tool	Description
bash	Execute shell commands with timeout and error handling
file_read	Read file contents with offset/limit support
file_write	Create or overwrite files (auto-creates directories)
file_edit	Surgical string replacement in files
grep	Regex search (uses ripgrep when available)
glob	Find files by pattern
git_diff	Show working tree, staged, or range changes
git_log	Recent commit history
git_commit	Stage and commit files
git_branch	Create, switch, or list branches
web_fetch	Fetch a URL (SSRF-protected)
ask_user	Ask the user a question
think	Extended reasoning scratchpad (no side effects)
rag_search	Semantic search over indexed documents
browser_open	Open a URL in a browser (requires --browser)
browser_screenshot	Take a screenshot of the current page
browser_click	Click an element by CSS selector
browser_fill	Fill an input field
browser_text	Extract text from the page

Effort Levels

Control how much reasoning the model applies:

Level	Thinking	Temperature	Description
low	Off	0.3	Fast responses, minimal reasoning
medium	Off	0.7	Balanced (default)
high	On	0.7	Extended thinking enabled
max	On	1.0	Maximum reasoning with higher token budget

cmdr --effort max "design a caching layer"
cmdr --fast "rename x to count"

Vision

Attach images to prompts for multimodal analysis:

# CLI flag
cmdr --image screenshot.png "what's wrong with this UI?"

# In REPL
/image path/to/diagram.png
> explain this architecture

Supported by Ollama (images field), OpenAI (image_url), and Anthropic (image content blocks).

---

Next: Multi-Agent Teams →