laila-cli
v1.0.5
Published
Laila: An advanced, local-first AI developer assistant with MCP, Vision, and Multi-Agent Swarm capabilities.
Maintainers
Readme
Laila (Advanced AI Developer OS)
Laila is an elite, local-first AI developer assistant that runs entirely in your terminal. Originally architected for absolute privacy (via Ollama) and maximum speed, Laila has evolved into a complete Agentic Operating System.
🚀 Core Features
- Hybrid Intelligence: Runs
llama3locally for free/private tasks, and gracefully scales to Anthropic/Gemini for heavy reasoning. - Multi-Agent Swarm (
/swarm): Spawn concurrent background sub-agents to solve complex, multi-step tasks in parallel. - Universal MCP Connectors (
/mcp): Natively plugs into your enterprise data (Postgres, GitHub, Slack, Jira, AWS, Google Drive) via the Model Context Protocol. - Visual QA & Computer Use (
/browse): Natively hooks into Playwright to navigate to your localhost, take UI screenshots, and visually QA your frontend code changes. - Semantic Project RAG (
/scan): Builds an instant, in-memory Knowledge Graph of your workspace using AST parsing—no heavy vector DBs required. - Self-Healing Build Loops: If your code fails to compile, Laila intercepts the raw compiler
stderrand automatically re-writes the fix. - Episodic Memory (
/remember): A global memory graph that remembers your precise architectural preferences across every project you touch. - Sandboxed Execution (
/sandbox): Writes and executes code natively, with dangerous commands automatically routed through disposable Docker containers for absolute host security.
📦 Installation
npm install -g laila-cli🎮 Quickstart
Just type laila inside any directory to launch the OS.
# Connect to your Postgres Database
laila> /mcp npx @modelcontextprotocol/server-postgres postgresql://localhost/mydb
# Trigger a Multi-Agent Swarm
laila> /swarm Build the frontend | Write the backend API | Setup Docker
# Visual UI Debugging
laila> /browse http://localhost:3000🧠 The "Ponytail" Philosophy
Laila is built on the "Ponytail" architectural philosophy: zero bloat, maximum stdlib usage, and brutal efficiency. She doesn't rely on 500MB of heavy Python ML libraries; she is purely driven by modern TypeScript, raw CLI power, and extreme Agentic logic.
Developed by Anurag Kumar
