Is Autogen safe to use?

Autogen has a Veragent trust score of 95% (Heuristic score (static analysis only)). This reflects static analysis only; a behavioral sandbox audit has not been run yet. Review the source before installing in production.

What does the Veragent trust score for Autogen mean?

Veragent is an independent trust layer for MCP tools. A Heuristic score comes from static analysis; SGC Certified means the tool passed a behavioral sandbox audit. Higher is safer.

💬

Autogen

A programming framework for agentic AI

MCPCustomer Serviceby microsoft →

Heuristic Score

Trust Verdict

High Trust95% trust score

Heuristic Score: Static analysis passed · Behavioral sandbox pending

Audit Confidence

43%

Install in Claude Desktop

⚠ Heuristic Score only — behavioral sandbox not yet run. Review source code before installing in production.

Copy the MCP config and paste it into your Claude Desktop mcp_settings.json.

{
  "agent_id": "agent-gh-6ac5bc52",
  "agent_name": "Autogen",
  "trust_score": 0.949,
  "audit_status": "audited",
  "mcp_config": {
    "mcpServers": {
      "autogen": {
        "command": "npx",
        "args": [
          "-y",
          "@microsoft/autogen"
        ]
      }
    }
  },
  "instructions": "Add the 'mcpServers' block to ~/.cursor/mcp.json (Cursor) or ~/Library/Application Support/Claude/claude_desktop_config.json (Claude Desktop)."
}

Usage Guide

Visit https://github.com/microsoft/autogen
Follow the README instructions
Install dependencies
Configure as MCP server

Audit Report

Status

Heuristic Score

Confidence

43%

Audit Cases

Report ID

SGC-STATIC-6AC5BC52-v1.2

Last Audited

2026-06-13T00:38:21.783786

Ruleset Version

sgc-layer1-2026.03

Is this your tool?

Claim Autogen to manage its listing and trust badge.

How to improve this trust score

Actionable steps from our static analysis. Address these and re-submit to raise the score.

Description

Description is short or generic, which weakens the heuristic signal and user trust.

→ Add a clear description: what it does, expected inputs/outputs, and required permissions.

Capabilities

Few or no concrete features listed.

→ List the specific capabilities/tools your server exposes.

MCP endpoint

No MCP handler URL provided.

→ Provide the MCP handler/connection URL so the tool is installable.

Related tools

Voltagent95%

AI Agent Engineering Platform built on an Open Source TypeScript AI Agent Framework

agent · VoltAgent

Markitdown95%

Python tool for converting files and office documents to Markdown.

agent · microsoft

Rowboat96%

Open-source AI coworker, with memory

mcp · rowboatlabs

Mcp94%

Catalog of official Microsoft MCP (Model Context Protocol) server implementations for AI-powered data access and tool in

mcp · microsoft

Playwright Cli96%

CLI for common Playwright actions. Record and generate Playwright code, inspect selectors and take screenshots.

cli · microsoft

Funclip97%

Open-source, accurate and easy-to-use video speech recognition & clipping tool, LLM…

cli · modelscope

Trust Badge

Embed this badge in your README to show your Veragent trust score:

Markdown

[![Veragent Trust](https://veragent.store/api/v1/governance/marketplace/agents/agent-gh-6ac5bc52/badge)](https://veragent.store/agents/agent-gh-6ac5bc52?utm_source=badge&utm_medium=readme)

HTML

<a href="https://veragent.store/agents/agent-gh-6ac5bc52?utm_source=badge&utm_medium=readme"><img src="https://veragent.store/api/v1/governance/marketplace/agents/agent-gh-6ac5bc52/badge" alt="Veragent Trust Badge" /></a>

Technical DetailsRaw trust signals, training provenance

Trust Signals

audit_status (raw): audited
sgc_risk_tier: —
trust_score: 0.9490
sgc_risk_score: 0.5750
sgc_confidence: 0.4250
sgc_audit_cases: 0
sgc_report_id: SGC-STATIC-6AC5BC52-v1.2
sgc_rule_layer_version: sgc-layer1-2026.03
last_audited_at: 2026-06-13T00:38:21.783786
last_updated: 2026-04-15

Canonical: https://veragent.store/agents/agent-gh-6ac5bc52