examples

Osiris Examples

Example pipeline files and AI Operation Package (AIOP) exports demonstrating Osiris capabilities.

Pipeline Examples (OML Format)

• mysql_to_local_csv_all_tables.yaml - MySQL to CSV export pipeline
• mysql_to_supabase_all_tables.yaml - MySQL to Supabase migration pipeline
• mysql_duckdb_supabase_demo.yaml - MySQL → DuckDB → Supabase transformation pipeline (NEW)

AIOP Examples

• aiop-sample.json - Complete AI Operation Package with all layers
• run-card-sample.md - Human-readable run summary in Markdown

Usage

Pipeline Examples

These files show the OML v0.1.0 format. To generate your own pipelines:

# Activate virtual environment first
source .venv/bin/activate

# Initialize configuration
python osiris.py init

# Generate pipeline through conversation
python osiris.py chat

# Run example pipeline with dry-run
python osiris.py run examples/mysql_to_local_csv_all_tables.yaml --dry-run

MySQL → DuckDB → Supabase Demo

The mysql_duckdb_supabase_demo.yaml demonstrates data transformation using DuckDB:

Purpose: Extract movie data from MySQL, compute director statistics using DuckDB SQL, and write results to Supabase.

Quick Start:

# Using make target (recommended)
make demo-mysql-duckdb-supabase

# Or manually from testing_env
cd testing_env
python ../osiris.py compile ../docs/examples/mysql_duckdb_supabase_demo.yaml
python ../osiris.py run --last-compile

Expected Output:

• Extracts ~14 movies from MySQL
• Computes statistics per director (movie count, avg runtime, total box office, ROI)
• Creates/updates director_stats_demo table in Supabase with aggregated results

Running in E2B Sandbox:

# Requires E2B_API_KEY environment variable
make demo-mysql-duckdb-supabase-e2b

# Or manually
cd testing_env
python ../osiris.py compile ../docs/examples/mysql_duckdb_supabase_demo.yaml
python ../osiris.py run --last-compile --e2b --verbose

Troubleshooting:

• Missing connections: Ensure osiris_connections.yaml exists with MySQL and Supabase configs
• Authentication errors: Check MYSQL_PASSWORD and SUPABASE_SERVICE_ROLE_KEY environment variables
• Table exists error: The pipeline uses write_mode: replace to overwrite existing data
• DuckDB not found: The driver is now registered in both local and E2B execution paths

AIOP Examples

After running any pipeline, explore the AI Operation Package:

# Run a pipeline to generate AIOP
python osiris.py run examples/mysql_to_local_csv_all_tables.yaml

# View the generated AIOP
cat logs/aiop/latest.json | jq '.narrative.summary'

# Export specific run as Markdown
python osiris.py logs aiop --last --format md

# Compare with sample AIOP structure
jq 'keys' docs/examples/aiop-sample.json

Quick AIOP Walkthrough

1. Enable AIOP (default in new installs):

   python osiris.py init
   # Ensure aiop.enabled: true in osiris.yaml

2. Run any pipeline:

   python osiris.py run examples/mysql_to_local_csv_all_tables.yaml

3. Explore exports:

   # AI-friendly JSON export
   ls logs/aiop/
   
   # Human-readable summary
   cat logs/aiop/*_runcard.md
   
   # Check for truncation
   jq '.metadata.truncated' logs/aiop/latest.json

See the main README.md for complete setup instructions.