Add AgentLenz — AI agent cost optimization

[AdelElo13]

AgentLenz

URL: https://github.com/AdelElo13/agentlenz
PyPI: pip install agentlenz

Description

Cut your AI agent costs across every provider. Not just observability — optimization. AgentLenz traces agent behavior, detects waste, recommends cheaper model routing, and enforces per-task budgets.

Key Features

• Provider-agnostic: Works with Anthropic, OpenAI, and any LLM provider via SDK wrapping
• Cost tracing: Automatic token/cost tracking per task, per agent, per session
• Model routing optimization: Detects when expensive models are used for simple tasks and recommends cheaper alternatives
• Budget enforcement: Set per-task cost limits that kill runaway agent loops
• Dashboard: Next.js web UI for visualizing agent costs and optimization opportunities

Packages

• sdk/ — Python SDK (open-source, PyPI: agentlenz)
• backend/ — FastAPI API server
• dashboard/ — Next.js web dashboard

Quick Start

import agentlenz
import anthropic

agentlenz.init(api_key="alz_...")
client = agentlenz.wrap(anthropic.Anthropic())
# Use client as normal — AgentLenz tracks costs automatically

Category suggestion

Developer tools / Observability & Cost optimization

[AdelElo13]

Thanks @wangranm-a11y! Interesting angle — rendering costs are definitely a blind spot in most agent cost discussions. Everyone focuses on LLM token spend but forgets the compute layer.

Edge/client-side rendering for video avatars at <400ms latency sounds compelling. Will check out SpatialReal 👍