+ P

Integration · Evaluate

Evaluate for Microsoft AutoGen agents

Capture every ConversableAgent message and GroupChatManager turn from your Microsoft AutoGen agents — with the actual speaker preserved, not flattened into one shared thread.

Book a demo → Request access

TL;DR

Prefactor evaluates your Microsoft AutoGen agents — score outcome quality against the captured spans, track drift by comparing custom spans across versions and environments, and classify every run's risk on two axes: data sensitivity and action consequence.

Evaluate

pillar

TS + Py

official SDKs

Spans

LLM · tools · agents

Tokens

usage captured

How to add evaluate to Microsoft AutoGen

Install the SDK

Add prefactor-core to your environment.

Register & create spans

Spans flow to Prefactor

Structured trace data lands in Prefactor as your agent runs.

# pip install prefactor-core
import os
from prefactor_core import PrefactorCoreClient, PrefactorCoreConfig
from prefactor_http import HttpClientConfig

config = PrefactorCoreConfig(
    http_config=HttpClientConfig(
        api_url="https://app.prefactorai.com",
        api_token=os.environ["PREFACTOR_API_TOKEN"],
    )
)

client = PrefactorCoreClient(config)
await client.initialize()

# then instrument your Microsoft AutoGen run with spans — see docs.prefactor.ai

Shown with the Prefactor SDK — a first-class, working integration today.

How the Microsoft AutoGen integration actually works

In a group chat, every message technically routes through the GroupChatManager, which selects the next speaker and broadcasts to the rest of the group — Prefactor's spans preserve the actual last_speaker for each turn, not just "a message arrived", so the real conversation flow between agents stays reconstructable.
Custom reply functions registered via register_reply() are captured the same way as built-in replies — instrumenting an agent doesn't require you to only use AutoGen's default reply behaviour.
Beyond auto-captured spans, use withSpan to record any custom step you define — an API call, a quality check, a business action.

What Prefactor captures for Microsoft AutoGen agents

Outcome quality

Score agent outputs against your own criteria, task by task, using the continuous run history as the record to measure against.

Drift detection

Compare custom spans for the same operation across versions and environments — a model update or prompt change shows up as a difference between spans, before it shows up as a quality drop.

Risk classification

Every run scored on data sensitivity and action consequence, weighted into a Low / Medium / High / Critical classification.

Data tagging

Tag PII and other sensitive fields once in the schema; every run carrying that tag becomes searchable — the record enforcement acts on.

Trace anything — not just LLM calls

The SDK captures LLM, tool, and agent spans automatically. With withSpan you wrap any operation in your own span type — an API call, a database query, a quality check, a business action — each with its own payload and schema. It all flows into the same Observe, Evaluate, and cost views.

research_competitorquality_checkpricing_lookuprefund_decisiondoc_retrieval …anything you define

import { withSpan } from "@prefactor/core";

// Wrap ANY operation in a span you define — an API call, a quality
// check, a business action — with its own spanType, inputs and schema
await withSpan(
  {
    name: "research competitor",
    spanType: "research_competitor",
    inputs: { competitor },
  },
  async () => {
    const results = await search(competitor); // your tool / API calls
    return summarize(results);                // captured as one span
  },
);

Nest spans to capture business-level actions, and start with permissive schemas you tighten over time. Instrumentation strategy →

An example run, span by span

Illustrative — a single Microsoft AutoGen run as nested spans.

Trace · Microsoft AutoGen run

1.84s6 spans1,210 tokensok

invoke_agent 1840

llm: plan 440

tool: search_kb 330

retriever: vector_search 210

llm: synthesize 700

tool: send_reply 320

agentllm calltool callretriever

Illustrative example.

Manual logging vs DIY vs Prefactor

Capability	Manual	DIY OpenTelemetry	Prefactor
LLM / tool / agent spans	Hand-rolled	✓ build it	✓ via the SDK
Token usage captured per call	—	Build it	✓
Configurable capture & sampling	—	Partial	✓
Hosted Admin UI (agents, instances)	—	—	✓
Risk profiles & audit trail	—	—	✓

Microsoft AutoGen evaluate — FAQ

Does evaluation reuse the same Microsoft AutoGen integration?+

Yes — evaluation runs on the spans the SDK already records for Microsoft AutoGen, so there is one integration, not a separate one for evaluation.

What is quality drift and how is it detected for Microsoft AutoGen agents?+

Drift is a change in behaviour over time — after a model update, a prompt change, or a data shift. Because a custom span records the same Microsoft AutoGen operation every time it runs, this week's spans can be compared directly against last week's, or against a different version or environment.

How is risk scored for Microsoft AutoGen runs?+

On two axes — data sensitivity (what kind of data the run touched) and action consequence (what the agent did with it) — combined via configurable weights into a Low/Medium/High/Critical classification per run.

Can I tag and find PII across Microsoft AutoGen runs?+

Yes — tag the fields that matter once in your agent's data schema, and every run carrying that tag becomes searchable across your whole Microsoft AutoGen deployment, not just the run it was tagged in.

Do I need a dedicated package for Microsoft AutoGen?+

You can instrument Microsoft AutoGen today with the framework-agnostic prefactor-core SDK; a dedicated package can be added on request.

What does Prefactor capture from Microsoft AutoGen?+

Prefactor records conversable-agent messages, group-chat turns, tool calls and LLM calls as structured, timestamped spans — so every Microsoft AutoGen run is captured as trace data you can reconstruct, search and export end to end.

Does Prefactor add latency or change how Microsoft AutoGen runs?+

No. Observability capture is designed to stay off your agent's critical path, so it doesn't alter your Microsoft AutoGen logic or your users' responses. The only part that acts inline is the optional runtime guardrails you enable per agent — by design, so a high-risk or low-confidence action can be held for human approval before it executes.

Can I evaluate agents built with Microsoft AutoGen and catch regressions?+

Yes. Once runs are captured, eval suites score quality and groundedness on real traffic, drift detection flags behaviour changes after deployment, and versioned eval history catches regressions before they ship — the observe → evaluate → improve loop applied to your Microsoft AutoGen agents.

Keep going

See it on your Microsoft AutoGen agents

Book a 15-minute setup and our team gets you evaluate in production.

Book a demo → Talk to integrations