Creating Modules
Lola modules can be a single Agent Skill or a full AI Context Module. An Agent Skill is a standalone SKILL.md with optional supporting files. An AI Context Module is a superset - it wraps one or more skills alongside AGENTS.md, commands/, and mcp.json inside a module/ directory.
Initialize an AI Context Module
This creates the AI Context Module structure:
my-module/
module/
AGENTS.md # AI main spec
skills/
example-skill/
SKILL.md # Skill following agentskills.io
commands/
example-command.md
agents/
example-agent.md
mcps.json # MCP settings
Note
lola mod init currently creates only the AI Context Module pattern. Standalone Agent Skill initialization (lola skill init) is planned for a future release.
Edit the skill
Edit your skill's SKILL.md following the AgentSkills.io standard:
---
name: my-skill
description: When to use this skill
---
# My Skill
Instructions for the AI assistant...
Add supporting files
Each skill can have its own scripts/, reference/, and assets/ directories:
my-skill/
SKILL.md
scripts/ # Executable scripts
reference/ # Documentation
assets/ # Other supporting files
Reference them with relative paths in your SKILL.md:
Add an AGENTS.md
The AGENTS.md provides module-level context that applies across all skills in the module. This is what elevates a collection of skills into an AI Context Module.
Add slash commands
Slash commands are custom commands that can be invoked with /command-name in AI assistants. Claude Code, Cursor, and Gemini CLI all support them. Create markdown files in commands/:
---
description: Review a pull request
argument-hint: <pr-number>
---
Review PR #$ARGUMENTS and provide feedback.
Use $ARGUMENTS for all args or $1, $2 for positional. Lola automatically converts commands to each assistant's native format (markdown for Claude Code and Cursor, TOML for Gemini CLI).
Register and install
See Skill Format for details on the SKILL.md specification.