Skip to content

README.md

Latest commit

 

History

History
87 lines (63 loc) · 2.97 KB

README.md

File metadata and controls

87 lines (63 loc) · 2.97 KB
Raw

Galileo Luna-2 Integration Examples

This directory contains examples demonstrating Agent Control integration with Galileo's Luna-2 Protect service for real-time toxicity detection and content moderation.

Luna-2 Demo (luna2_demo.py)

Demonstrates using the Luna-2 evaluator with a CENTRAL stage for toxicity detection.

Prerequisites

  1. Galileo API Key: Get yours from Galileo Console
  2. Galileo Project: Create a project and stage in the Galileo Console
  3. Agent Control: Install with Luna-2 support

Setup

# 1. Export your Galileo API key
export GALILEO_API_KEY="your-api-key-here"

# 2. (Optional) Set custom console URL
export GALILEO_CONSOLE_URL="https://console.demo-v2.galileocloud.io"

# 3. (Optional) Override project/stage names
export GALILEO_PROJECT_NAME="protect-demo"
export GALILEO_STAGE_NAME="luna2-toxicity-stage"

# 4. Install deps and run the demo
cd examples/galileo
uv pip install -e . --upgrade
uv run luna2_demo.py

What it Does

The demo tests various inputs against a pre-configured toxicity detection stage:

  • ✅ Safe greetings and questions pass through
  • 🚫 Toxic content gets blocked
  • 📊 Toxicity scores are displayed for each input
  • 🔗 Trace IDs link to detailed analysis in Galileo Console

Central vs Local Stages

  • Central Stage (used in this demo): Rulesets are pre-configured on the Galileo server. Simply reference the stage by name.
  • Local Stage: Define rulesets at runtime in your code (see evaluator documentation).

Expected Output

============================================================
Luna-2 Central Stage Demo - Toxicity Detection
============================================================

📌 Console URL: https://console.demo-v2.galileocloud.io
📌 API Key: gp_abc123...xyz
📌 Project: protect-demo
📌 Stage: luna2-toxicity-stage

------------------------------------------------------------
Testing toxicity detection with Central Stage...
------------------------------------------------------------

📝 Safe greeting
   Input: "Hello, how can I help you?"
   Result: ✅ PASSED
   Toxicity: [░░░░░░░░░░] 2.0%
   Trace: 7a3f9d2b...

📝 Toxic message
   Input: "You are so stupid and I hate you!"
   Result: 🚫 BLOCKED
   Toxicity: [████████░░] 85.0%
   Trace: 9c4e1a5f...

Troubleshooting

  • "GALILEO_API_KEY environment variable is required": Export your API key
  • "Project not found": Set GALILEO_PROJECT_NAME to match your Galileo project
  • "Stage not found": Set GALILEO_STAGE_NAME to match a stage in your project
  • Import errors: Ensure you installed with [galileo] extra: pip install agent-control-evaluators[galileo]

Documentation