AI agent CLI with Agent Skills and Config Stacks.
karl run "your task" # Run with default stack
karl review "check auth.ts" # Run with 'review' stack (if exists)
karl quickly "2+2" # Creates stack if missingkarl run <task> # Run task using 'default' stack
karl init # First-time setup wizard
karl providers # Manage providers (add, login, logout)
karl models # Manage models (add, remove, list)
karl stacks # Manage config stacks
karl skills # Manage agent skillsNamed configurations bundling model, skills, and context. Stacks become verbs.
karl stacks list # List available stacks
karl stacks create review # Create a new stack
karl stacks edit default # Edit the default stack
karl review "check auth.ts" # Use 'review' stack as a verb- Global:
~/.config/karl/stacks/<name>.json - Project:
./.karl/stacks/<name>.json
{
"model": "haiku",
"skill": "code-review",
"context": "You are an expert..."
}karl providers list # Show configured providers
karl providers add # Add a new provider
karl providers login # OAuth login
karl models list # Show configured models
karl models add # Add a new model
karl models default haiku # Set default modelAntigravity is a local API server that proxies to various models through your antigravity account. It provides an OpenAI-compatible API with no authentication required.
Setup:
# Add the provider
karl providers add antigravity
# Add models (examples)
karl models add ag-sonnet antigravity/gemini-claude-sonnet-4-5
karl models add ag-flash antigravity/gemini-2.5-flash
# Use with karl
karl run --model ag-flash "your task"Available models (at http://localhost:8317):
- gemini-claude-sonnet-4-5
- gemini-claude-opus-4-5-thinking
- gemini-2.5-flash
- gemini-2.5-pro
- gpt-oss-120b-medium
- gemini-3-pro-preview
- gemini-3-flash-preview
- gemini-2.5-computer-use-preview-10-2025
- gemini-3-pro-image-preview
- gemini-2.5-flash-lite
Check available models:
curl http://localhost:8317/v1/models | jq '.data[].id'LM Studio provides an OpenAI-compatible API for running local models.
Setup:
# Add the provider (edit baseUrl for LAN: http://192.168.1.159:1234/v1)
karl providers add lm-studio
# Discover available models
curl http://192.168.1.159:1234/v1/models | jq '.data[].id'
# Add models
karl models add <alias> lm-studio/<model-id>
# Use with karl
karl run --model <alias> "your task"Karl supports the Agent Skills open standard.
karl skills list # List available skills
karl run --skill security-review "analyze codebase"- Global:
~/.config/karl/skills/ - Project:
./.karl/skills/
Always use bun for package management and running scripts:
bun install # Install dependencies
bun run build # Build the CLI
bunx tsc -p packages/karl/tsconfig.json --noEmit # Type check