traeco-sdk 1.0.2

AI agent cost intelligence — trace LLM calls, attribute costs, get optimization recommendations.

traeco-sdk

Stop paying for AI agent waste. Traeco traces every LLM call, attributes costs per function, and surfaces exactly where to cut spend.

Install

pip install traeco-sdk

Quickstart

import traeco
from anthropic import Anthropic

traeco.init(
    api_key="tk_live_...",
    host="https://backend-kynarochlani-4185s-projects.vercel.app",
    agent_name="my-agent",
)

client = traeco.wrap(Anthropic())

# Your existing code — unchanged
response = client.messages.create(model="claude-sonnet-4-6", ...)

Per-function cost breakdown

@traeco.span("market_analysis")
def analyze(data):
    return client.messages.create(...)

@traeco.span("trade_decision")
def decide(analysis):
    return client.messages.create(...)

Each @span becomes its own row in the dashboard — cost, tokens, latency, and recommendations per function.

Works with OpenAI too

from openai import OpenAI
client = traeco.wrap(OpenAI())

Environment variable

import os
traeco.init(api_key=os.getenv("TRAECO_API_KEY"), ...)

Full docs

See DOCS.md for the complete reference including Claude Code integration prompts.