Documentation

All ElectriPy documentation lives on GitHub. Start with your use case, then move into packages, recipes, API references, and architecture guides.

Open Full Docs on GitHub

Start here

Choose the path that matches your use case.

I want observability and tracingI want governed executionI want model routing and fallbacksI want evaluation and regression checksI want MCP integrationI want reusable skillsI want realtime orchestration

Getting Started

Installation

Install ElectriPy via pip and verify your environment.

Quickstart

Get a working observable agent flow running in under 5 minutes.

Architecture Overview

How the seven flagship packages compose into observable agent systems.

Flagship Packages

electripy_observe

Observability layer — tracing, redaction, OpenTelemetry adapters, span helpers.

electripy_mcp

Typed MCP client/server toolkit — transport abstraction, tool registration, capability discovery.

electripy_router

Routing engine — model/provider selection, capability matching, fallback chains.

electripy_evals

Evaluation framework — scorers, baselines, CI-friendly reports, retrieval metrics.

electripy_policy

Runtime policy engine — action gating, approvals, evidence checks, audit-ready decisions.

electripy_skills

Skill packaging — manifests, asset resolution, validation, portable skill units.

electripy_realtime

Realtime orchestration — session state, event pipeline, cancellation, streaming.

User Guide

CLI Guide

Build and ship typed CLI tools using the ElectriPy CLI framework.

AI Product Engineering

Patterns and primitives for production AI product development.

Policy Gateway (Legacy)

Enforce content policies and guardrails on LLM traffic via the policy gateway.

RAG Eval Runner

Evaluate retrieval-augmented generation quality systematically.

Reference

API Reference

Full API reference for all ElectriPy modules and components.

Component Maturity Model

Understand the stability and readiness level of each component.

Project

Changelog

Release notes and version history for ElectriPy.

Contributing

How to contribute code, docs, and feedback to the project.

Security

Security policy and how to report vulnerabilities.

All documentation is maintained in the GitHub repository alongside the source code.

github.com/inference-stack-llc/electripy-ai/docs