Lookio CLI

The official Command Line Interface for Lookio—the API-first Knowledge Retrieval (RAG) platform.

Turn your documents into "Automated Expertise" that can be queried directly from your terminal or integrated seamlessly into AI agent workflows.

Installation

Install the Lookio CLI globally using npm:

npm install -g lookio-cli

Alternatively, you can run commands safely without installing using npx lookio-cli <command>.

Getting Started

1. Authentication

To use the CLI, you must first authenticate with your Lookio API key.

lookio login "YOUR_API_KEY"

2. Run a Query

Query your knowledge base through a specific persona (Assistant). Lookio supports multiple query modes (eco, flash, europe, deep) to balance cost, speed, and intelligence.

lookio query "What is your return policy?" --assistant "ASSISTANT_ID" --mode flash

Advanced Usage

Managing Resources

Resources are your raw data (PDFs, TXTs, live URLs, or raw text blocks).

# List all resources
lookio resources list

# Upload a local document (uses secure multipart/form-data)
lookio resources add --title "Q4 Report" --file "./q4-financials.pdf"

# Ingest a live webpage
lookio resources add --title "Blog Post" --url "https://example.com/blog/1"

# Delete a resource
lookio resources delete "RESOURCE_ID"

Managing Assistants

Assistants are custom AI personas with specific context and output guidelines.

# List all assistants
lookio assistants list

# Create a new assistant
lookio assistants create \
  --name "HR Bot" \
  --context "You are a helpful HR assistant..." \
  --guidelines "Answer in bullet points." \
  --type "All resources"

# Update an existing assistant's instructions
lookio assistants edit "ASSISTANT_ID" --name "Senior HR Bot"

# Delete an assistant
lookio assistants delete "ASSISTANT_ID"

Built for Vibe Coding & AI Agents

This CLI is designed to be highly token-efficient, allowing autonomous AI agents (like Cursor, GitHub Copilot, or Claude Desktop) to interact directly with your Lookio knowledge base, create assistants dynamically, and pipeline RAG workflows straight from the terminal.

License

MIT