PyStreamMCP

Two-stage selective intelligence: metadata-first filtering + contextual reranking. 90-95% token reduction. OpenAI-compatible MCP orchestration layer for streaming.

Latest Version: 1.1.1

Features

• ✅ 90-95% token reduction (selective intelligence)
• ✅ 8+ connected tools with latency tracking
• ✅ Metadata-first filtering & contextual reranking
• ✅ Production-ready CLI dashboards
• ✅ Keyboard shortcuts for quick access
• ✅ OpenTelemetry support (6 backends)

Installation

pip install pystreammcp

Quick Start

# Setup shortcuts
bash scripts/setup_shortcuts.sh

# View dashboard
dash-pystreammcp              # Static snapshot
dash-pystreammcp-live         # Live monitoring
dash-pystreammcp-export       # Export metrics

# Connect tools
pystreammcp list-tools
pystreammcp connect --provider openai

Dashboard

Real-time MCP orchestration metrics:

• dash-pystreammcp - View tool status, selective intelligence stats
• dash-pystreammcp-live - Watch active executions, routing performance
• dash-pystreammcp-export - Export to JSON

Metrics: Status, Connected Tools (8+), Selective Intelligence Reduction (90-95%), Active Jobs, Routing Performance

OpenTelemetry

Export to Prometheus, Datadog, Honeycomb, New Relic, Jaeger, or X-Ray.

export OTEL_EXPORTER_OTLP_PROTOCOL=prometheus
dash-pystreammcp-live

See OTEL_SETUP_GUIDE.md.

Production Deployment

Kubernetes, Docker Compose, and standalone patterns included.

See PRODUCTION_DEPLOYMENT.md.

Documentation

• DASHBOARD_SHORTCUTS.md - Keyboard shortcuts
• OTEL_SETUP_GUIDE.md - OpenTelemetry setup
• PRODUCTION_DEPLOYMENT.md - Deployment patterns

Repository

• GitHub: https://github.com/Mullassery/PyStreamMCP
• PyPI: https://pypi.org/project/pystreammcp

License

MIT