Manifest | The Open Source LLM Router for OpenClaw | WarmIndex

Manifest is an open-source LLM router plugin for OpenClaw, the self-hosted AI agent, that intelligently routes each query to the most cost-effective model, saving users up to 70% on AI inference costs.

🚀 Overview

Manifest is a smart routing layer for OpenClaw, an open-source personal AI assistant that runs locally and handles tasks like email, calendar, and more via messaging apps. It analyzes every incoming query in under 2ms using a 23-dimension scoring algorithm to select the optimal LLM model based on complexity, cost, and performance, avoiding the inefficiency of sending simple tasks to expensive large models.

The project solves the problem of over-provisioning in OpenClaw setups, where all requests default to the same model, leading to wasted tokens and higher costs. It's built for developers, AI enthusiasts, and teams running self-hosted AI agents who want to optimize expenses, track usage, and maintain privacy without coding changes.

Creators built Manifest to make local AI more economical and observable, with full open-source transparency and options for both local and cloud deployment.

✨ Key Features

Intelligent Routing: Analyzes queries locally (<2ms) and routes to the best model across 300+ options from multiple providers, cutting costs by up to 70%.
Cost Tracking & Alerts: Real-time dashboard tracks tokens, costs, and usage; set budgets and get alerts via OpenTelemetry.
Privacy-First: Local mode keeps all data on-device; cloud is a blind proxy with no access to prompts.
Automatic Fallbacks: Switches to backup models if a provider fails.
Easy Install: One-command OpenClaw plugin; no coding needed. Upcoming: More provider integrations.