# Sonatype Guide steers AI coding assistants to safer open source and automates dependency upkeep

> Sonatype Guide steers coding assistants to safer packages and curbs bogus dependency picks. Early pilots saw 300% better security and 5x lower upgrade effort.

- Type: AI news
- Page: https://completeaitraining.com/news/sonatype-guide-steers-ai-coding-assistants-to-safer-open/
- Published: 2025-12-10
- Updated: 2025-12-10
- Author: Jeroen Erne
- Jobs: IT and development
- Reading time: 2 min

### Updated 06:00 EST / December 09, 2025 - Sonatype launches Guide to secure AI-assisted development

Sonatype has introduced Sonatype Guide, a developer tool built to make AI-assisted software development faster, safer, and more efficient. It serves as an intelligent backbone that steers coding assistants toward secure, high-quality open-source components and keeps dependencies healthy over time.

#### Why this matters

AI coding assistants are trained on public data that can be months or years out of date. The result: they often recommend vulnerable, low-quality, or even imaginary packages. Sonatype's upcoming study reports that leading LLMs hallucinate packages up to 27% of the time-wasting tokens, creating rework, delaying delivery, and adding avoidable security risk.

#### Early results from enterprise testing

Pre-launch users reported more than a 300% improvement in security outcomes while reducing total security remediation. They also cut dependency-upgrade costs by more than 5x versus the leading competitive approach, measured in both spend and developer hours.

#### How Guide fits into your workflow

Guide works alongside popular AI coding assistants-GitHub Copilot, Google Antigravity, Claude Code, Windsurf, IntelliJ with Junie, Kiro from AWS, and Cursor-so teams can keep existing habits while raising the bar on dependency quality and safety.

#### Key capabilities

- **MCP server for coding assistants:** Intercepts package recommendations in real time and routes developers to secure, reliable versions before code hits the repo. Learn more about the Model Context Protocol [here](https://github.com/modelcontextprotocol/specification?ref=completeaitraining.com).
- **Enhanced OSS search:** Instant, trustworthy package decisions without hunting across docs and registries.
- **Enterprise-grade API:** Complete, unrestricted, backward-compatible access to reliable component data.
- **Built on Sonatype Intelligence:** Real-time signals on open-source quality, security, and project health that flag vulnerabilities, deprecations, and malicious packages early.

#### What Sonatype is saying

CEO Bhagwat Swaroop said organizations want AI-driven productivity without giving up security or long-term maintainability. He described Guide as bringing discipline and intelligence to AI-assisted development by steering assistants to secure, reliable components and automating dependency work that slows teams down-signaling a meaningful step forward for customers and the industry.

#### What this means for engineering leaders

- Cut noise from AI-generated dependency suggestions and standardize on safe defaults.
- Lower supply chain risk by blocking malicious or deprecated packages before they land.
- Reduce token waste from hallucinated packages and shrink remediation backlog.
- Improve long-term maintainability with automated dependency hygiene.
- Back decisions with auditable, enterprise-grade data and policy guardrails.

#### Practical next steps

- Start with a pilot repo and wire Guide into your existing coding assistant.
- Define policies for allowed package sources, versions, and risk thresholds.
- Track metrics: vuln count per PR, time-to-fix, token spend, and upgrade effort.
- Roll out to more teams once you prove reduced rework and security wins.

If you're formalizing AI coding across your org, consider upskilling your team on secure AI-assisted workflows. A good place to start is this focused pathway: [AI Certification for Coding](https://completeaitraining.com/certification/12b-ai-certification-for-coding/).

For broader context on supply chain risk scoring, see [OpenSSF Scorecards](https://securityscorecards.dev/?ref=completeaitraining.com).

---

This is the Markdown form of https://completeaitraining.com/news/sonatype-guide-steers-ai-coding-assistants-to-safer-open/. The whole site is mapped at https://completeaitraining.com/llms.txt.
