Build CLIs

Build typed command-line tools that also expose MCP tools and llms.txt discovery.

1 min read 106 words

Build human-facing commands and agent-facing tools from the same typed Python functions. Start with commands and groups, then add schemas, context, output formatting, llms.txt, and MCP.

Commands
Register typed command handlers and run them from argv
Groups
Organize commands into nested namespaces
Context and Output
Use injected context, global options, and structured returns
MCP Server
Expose commands as MCP tools and run the gateway
CLI & Commands
AI-native CLI with typed commands, automatic argparse, MCP tools, and llms.txt.
Output Formatting
Structured output with plain, JSON, table, and template formats.
Command Groups
Nested command groups for hierarchical CLI structures.
Context & Global Options
Execution context with verbosity levels, output format, and custom global options.
Lazy Loading
Deferred command imports for fast CLI startup with large command sets.
MCP Server
Expose CLI commands as MCP tools, register CLIs for discovery, and run a unified gateway.
llms.txt Generation
Generate AI discovery documents from CLI command definitions.
Help Rendering
Drop-in argparse HelpFormatter using Kida templates.

In This Section

CLI & Commands AI-native CLI with typed commands, automatic argparse, MCP tools, and llms.txt.
Command Groups Nested command groups for hierarchical CLI structures.
Help Rendering Drop-in argparse HelpFormatter using Kida templates.
Lazy Loading Deferred command imports for fast CLI startup with large command sets.
MCP Server Expose CLI commands as MCP tools, register CLIs for discovery, and run a unified gateway.
Related Pages
Dispatch Contract How CLI, programmatic, and MCP command calls agree. Related
Quickstart Build a typed CLI command that also works as an MCP tool. Related
From Python Fire Narrow automatic Python object exposure into explicit Milo command contracts. Related