Skip to main content

Changelog

Follow new updates and improvements to CybeDefend.

CybeDefend v1.3.0

MCP Check and Google Antigravity support in VibeDefend, a new CodeMap in AppSec, and simpler pricing across both products.

Now that VibeDefend and AppSec are standalone products, this changelog is split by product.

VibeDefend

🚀 New Features

I. MCP Check

VibeDefend now inspects every MCP server installed or used across all your AI coding agents, and classifies each one as malicious, suspicious, or safe. Admins can ban an MCP server for the entire organization: the ban is enforced directly in every user’s agent sessions, with no action needed on their side.

II. Google Antigravity support

VibeDefend now supports Google Antigravity, alongside Claude Code, Codex, Cursor, Copilot, and Windsurf.

III. Usage-based billing

VibeDefend now uses usage-based billing by default, so you only pay for what your team actually uses. Enterprise customers get a custom quote with fixed pricing.

✨ Improvements

Action Guard false negatives are cut in half. Action Guard now also analyses the scripts your agent writes and then decides to execute during a session, so a dangerous action can no longer slip through by being wrapped in a script file.

AppSec

🚀 New Features

I. CodeMap: routes, guards, and reachability

As announced in v1.1.3, the new CodeMap identifies your application’s routes and the guards protecting them, and shows whether a finding can actually be reached from a route. This lays the groundwork for what comes next: route reachability will feed the priority score and the exploitability metrics of the CVSS 4.0 score.

II. Simplified pricing

Choosing a plan now takes a single step: set your number of developers and projects with the slider, and the right plan is calculated for you. The Developer and Scale plans are retired and replaced by two plans:

  • Pro: from 1 to 5 developers, priced to your team size.

  • Team: from 6 to 25 developers, adding reports, security policies, SBOM / AI-BOM, EU AI Act compliance, and Jira integrations.

🛠️ Bug Fixes

Fixed the scrollbar in code snippets in AppSec.

CybeDefend v1.2.0

VibeDefend becomes a standalone product, 180+ new security rules for AI coding agents, and a faster, near-perfect Action Guard.

🚀 New Features

I. VibeDefend and AppSec, now two standalone products

VibeDefend and the CybeDefend AppSec platform are now fully separated. Each product has its own interface and its own billing, and an AppSec subscription is no longer required to use VibeDefend. Secure your AI coding agents, your codebase, or both: you only pay for what you need.

II. 180+ new VibeDefend security rules

The VibeDefend ruleset grows by more than 180 security rules, with new categories:

  • AI & LLM Applications

  • Cryptography

  • Mobile Applications

  • Kubernetes & Docker Compose

  • GitHub Actions

Coverage spans Python, JavaScript, TypeScript, Java, Go, C#, PHP, Ruby, Kotlin, and Rust.

✨ Improvements

Rules delivered to your agent are 31.2% more relevant. While your agent codes, VibeDefend now picks the security and business rules that actually apply to the file being edited more accurately, and rules that should have been delivered but were missed drop by 72%. In return, delivery latency rises slightly, to 250 to 350 ms per file edit.

Action Guard is faster and more precise. Each action is now evaluated in around 350 ms, and precision rises from 92.7% to 99.9%, so legitimate actions go through and dangerous ones stay blocked.

🛠️ Bug Fixes

Fixed API response times that could degrade during sudden load spikes.

Fixed a bug affecting Cybe AutoFix.

CybeDefend v1.1.3

Cybe Analysis runs six times faster, a new organization-wide findings API, and a round of fixes across notifications, container scans, and AutoFix.


🚀 New Features

I. Organization-wide findings API

New endpoint GET /organization/{organizationId}/findings lists security findings across an entire organization in a single call, instead of querying project by project. Built for GRC and SIEM exports, so your findings can feed your existing compliance and monitoring pipelines.

Note that a restrictive rate limit of 10 requests per minute is in place on this endpoint.

✨ Improvements

Cybe Analysis is now 6x faster at analysing vulnerabilities, cutting the time between a scan finishing and an audited, prioritised result.

Cybe Analysis is also more stable on large codebases, which now go through without degradation. This work lays the groundwork for the new parsing engine coming next: endpoints will be identifiable directly in the graph, and you will be able to see which execution paths a given vulnerability actually reaches.

🛠️ Bug Fixes

  • Fixed notification preferences in /profile that were ignored for zero-day and scan-complete emails, which kept being sent after being turned off.

  • Fixed a rare infinite loading state on the dashboard when a user had many tabs open in parallel.

  • Fixed false positives in VibeDefend's Action Guard, which was blocking some harmless actions.

  • Fixed GitHub project renames not being propagated to CybeDefend; a renamed repository now stays in sync.

  • Fixed container scans failing on ARM images.

  • Fixed SCA AutoFix that did not run on some projects.

  • Fixed code parsing that failed to trigger when a scan was queued immediately before it.

Earlier updates