Snyk Blog In this article Cloud-side AI inventory across Claude Enterprise Discovery: Every Claude asset in your inventory Risk Assessment: Per model risk signals across key categories that matter Compliance: Tool-level permission visibility, with the audit trail How it works Where this fits in the broader Snyk + Anthropic picture Get started today Snyk announces Anthropic updates: Evo integrates with Claude Enterprise, and Snyk Desk comes to Claude Desktop Written by Ranko Cupovic May 21, 2026 0 mins read Today, we're announcing two new integrations with Anthropic that cover both sides of AI-assisted development. Evo by Snyk now integrates with Anthropic's Claude Enterprise, giving security and compliance teams a complete inventory of their Claude environment models, approved MCP servers, per model risk signals, and tool-level permissions in the platform they already use to govern the rest of the stack. At the same time, the Snyk Security Desktop Extension is now available in Claude for macOS and Windows, embedding real-time scanning and vulnerability context directly into the developer workflow. For the past decade, security and compliance teams have built mature programs around their SaaS estate. This has allowed teams to know who has access to what in Microsoft 365, which apps are connected to your Slack workspace, and which AWS roles can touch production. The discipline is settled: discover the assets, score the risk, and prove the controls to your auditor. Cloud-side AI inventory across Claude Enterprise Evo by Snyk, the industry’s first AI-native security orchestrator, now integrates with Anthropic Claude Enterprise . Available in early access for Snyk Evo customers starting today, the integration uses the Claude Compliance API to pull a snapshot of your Claude Enterprise environment directly into the Evo alongside the code, machines, packages, and cloud surfaces you already monitor. Play Video: Govern Claude in Evo Evo gives security teams an inventory of AI assets across three surfaces: what's in their code, what's running on their developers' machines, and what's deployed through cloud AI platforms. Today's release specifically adds the third surface: an integration with the Claude Compliance API that offers an org's cloud-side view of the Claude Enterprise configuration, models, MCP servers, and tool permissions. The integration helps deliver three of the Evo AI-SPM pillars you already use for the rest of your stack: Discovery, Risk Assessment, and Compliance. Discovery: Every Claude asset in your inventory Every Claude model your organization has used surfaces as an Evo asset, with usage breakdowns across chats, projects, organizations, and users. Every MCP server your Claude Enterprise admins have approved appears alongside them, with the projects (public and private) it's connected to and the users using it. This means no more screenshot threads and no more "let me ask the Claude admin and get back to you." Risk Assessment: Per model risk signals across key categories that matter Once a model is in your inventory, Evo enriches it with the risk profile maintained in a continuously tested database of model risk signals across the categories that drive real incidents: bias and discrimination, insecure code generation, sensitive data exposure, attack reconnaissance, and safety guardrail bypass. You see one score per category on a 0–1000 risk index, you can drill into the test results behind each one, and you can prioritize the way you already prioritize the rest of your inventory. The Compliance API integration puts the model in your inventory, and Evo tells you what's risky about it. Compliance: Tool-level permission visibility, with the audit trail Because Evo already sees what's running on your developer machines, the Compliance API integration unlocks something neither side can do alone: we can detect drift between the MCP tool permissions your admins approved in Claude Enterprise and what your developers are actually running locally. This is the piece your GRC team will care about most. For every MCP server, Snyk surfaces the tool-level permission ceilings your Claude Enterprise admins have set — delete_task marked as Restricted, read_inventory marked as Read-only, and so on, along with who set the restriction and when. That's an auditor artifact, not a screenshot. When your next compliance review asks, "How do you enforce least privilege on AI agent tools?", you have an answer with a date and a name on it. How it works Setup is a two-step flow. Your Claude Enterprise Primary Owner generates a Compliance Access Key in Claude Enterprise under Settings → Data and Privacy, with three read-only scopes (read:compliance_activities, read:compliance_user_data, read:compliance_org_data). Paste that key into Evo, click Ingest snapshot, and Snyk calls the Anthropic Compliance API to pull in organizations, users, projects, chats, models, and MCP servers, all as native Evo assets you can search, score, and govern. Ingestion is on-demand, so you can re-run it whenever you want a fresh snapshot. Also new: Snyk Security in Claude Desktop While Evo gives security teams a cloud-side view of their Claude Enterprise environment, developers need security context when code is written. That's why we're also announcing that the Snyk Desktop Extension is now available in Anthropic's directory of extensions for Claude Desktop on macOS and Windows. Play Video: Connect Snyk to Claude Built on the Model Context Protocol (MCP), the open standard introduced by Anthropic, the extension embeds Snyk's scanning and vulnerability context directly into the Claude developer workflow. As Claude generates or edits code, Snyk provides real-time, actionable findings so developers can catch issues at inception, not after they've shipped. Connect a repository and Claude and Snyk will work together to scan existing code, prioritize what matters, and surface developer-ready fixes for the backlog you already have. AI assistants are transforming how software gets written, and research consistently shows that a significant share of AI-generated code introduces security vulnerabilities. Traditional AppSec was built for humans writing code over weeks; " secure at inception " is built for the new reality of agentic development , where thousands of lines can be generated in seconds. Together, the desktop extension and the Evo + Claude Enterprise integration give organizations both sides of the equation: developers get guardrails where code is written, and security teams get governance over how Claude is deployed across the enterprise. Where this fits in the broader Snyk + Anthropic picture This Evo integration with Claude Enterprise via the Claude Compliance API is the latest milestone in how we work with Claude. Earlier this month, we announced that Claude's frontier reasoning capabilities are being brought directly into the Snyk AI Security Platform , powering automated vulnerability discovery, prioritization, and developer-ready fixes across code, dependencies, containers, and AI-generated artifacts. With the Snyk Security Desktop Extension now available in Claude for macOS and Windows, we’re putting real-time scanning and vulnerability context into the hands of every developer using Claude to write code. Taken together, these integrations close the loop across the full lifecycle of AI-assisted development. Claude makes Snyk smarter at fixing vulnerabilities inside the AI Security Platform. Snyk makes Claude safer at the developer's desktop, catching issues at inception through the MCP-powered extension. Evo by Snyk makes Claude governable at the enterprise level, giving security and compliance teams the inventory, risk signals, and audit trail they need wherever Claude is deployed. Get started today The Evo integration with Claude Enterprise is available today in early access for Evo customers. To enable it for your tenant, reach out to your Snyk account team. To learn more: Read the integration setup guide See Anthropic's launch announcement Request a demo of Evo AI-SPM You can’t govern AI you can’t see Start with Discovery. Start with Evo AI-SPM. Uncover every AI component hidden in your codebase and apply organization-wide governance. Book a Demo