Back to articles
June 3, 2026

Ammar Askar Leaks VS Code OAuth Exploit Amid Microsoft Disclosure Revolt

Independent researcher Ammar Askar publishes a working Visual Studio Code exploit in roughly one hour, citing dismissed MSRC submissions and escalating the zero-day conflict driven by Nightmare Eclipse.

Close-up photo of hands typing on a keyboard with code visible on a blurred background screen.Photo: Luke Lung / Unsplash

Independent researcher Ammar Askar bypassed Microsoft's coordinated disclosure protocol to publish a working Visual Studio Code exploit in roughly one hour, citing a pattern of dismissive treatment from the Microsoft Security Response Center (MSRC). The flaw allows an attacker to harvest unscoped OAuth tokens and gain read-write access to linked GitHub repositories.

Askar's rapid publication mirrors the tactics associated with "Nightmare Eclipse," a suspected former Microsoft employee who has already released six Microsoft/Windows zero-days—including three actively exploited flaws—since early 2026. The latest incident signals a breakdown in the voluntary disclosure framework that typically separates researchers from vendors.

The github.dev attack chain

The exploit targets the integration between Visual Studio Code and its cloud-hosted workspace, github.dev. The attack relies on abusing the Workspace Recommendations feature to inject a malicious extension into the user's environment.

Execution requires minimal interaction from the victim. A user simply opens a Jupyter Notebook file (.ipynb) via github.dev. Behind the scenes, hidden HTML and JavaScript execute within the browser-based Webview. The script simulates a keyboard shortcut sequence engineered to force-approve the malicious extension popup automatically.

Once installed, the extension extracts unscoped OAuth tokens from the session. Because these tokens lack scope limitations, they grant the attacker full read and write capabilities across every public and private repository connected to the compromised account.

This mechanism demonstrates how trusted developer toolchains can serve as initial footholds. The extension recommendation system transforms a routine notebook view into a credential-scraping vector, expanding the effective supply-chain attack surface beyond traditional distribution channels.

The dispute over disclosure norms

Askar's decision to go public stems from repeated friction with MSRC. According to his account, a prior submission was silently patched, stripped of attribution, and ultimately classified as "not having any security impact." Similarly, a cross-site scripting vulnerability identified by Starlabs was downgraded to low severity or ruled ineligible.

Rather than wait for a resolution, Askar contacted an "old" associate at VS Code and then published the proof-of-concept roughly sixty minutes later.

Microsoft responded swiftly. The MSRC blog condemned the uncoordinated dump, specifically referencing the Digital Crimes Unit. The post framed the leak as a violation of responsible disclosure practices, implying potential legal consequences for such actions.

Industry reaction has been hostile toward Microsoft's framing. Critics argue that invoking law enforcement resources against independent researchers violates long-standing norms and actively discourages qualified contributors from engaging with the program. The backlash echoes the controversy surrounding earlier zero-day leaks attributed to the Nightmare Eclipse collective.

That group continues to operate aggressively. Following the latest dump, Nightmare Eclipse has issued warnings of additional disruptive releases scheduled for July 14, 2026, contingent on whether Microsoft addresses underlying grievances.

Our read

The patching window has effectively collapsed. Enterprises relying on Microsoft tooling now face detection and blocking mandates measured in hours rather than the standard twelve-to-fourteen-day grace period. When vendors treat qualified reporters as adversaries, the timeline shrinks until there is no operational safety net left.

Microsoft's reliance on the Digital Crimes Unit creates a perverse incentive structure. By signaling that uncoordinated disclosure invites prosecution, the company forces researchers into silence or rebellion. Both outcomes degrade visibility. Silence hides flaws; rebellion ensures they are weaponized by anyone willing to scrape the public feed.

The technical vector reinforces this strategic error. The ability to hijack OAuth tokens through a notebook viewer proves that the boundary between "trusted tool" and "attack surface" has dissolved. Developers assume extensions recommended by their host environment carry implicit vetting. That assumption no longer holds.

The upcoming July 14 deadline serves as a pressure test. Whether the next wave comes from Nightmare Eclipse or a new cohort of aggrieved hunters depends less on individual malice and more on Microsoft's willingness to repair the feedback loop. Until the vendor treats disclosure as a collaborative engineering discipline rather than a compliance hurdle, critical flaws will remain in the wild until they are extracted by force.


Reporting from The Register and TechCrunch.

The Signal

AI-generated brief

Microsoft’s adversarial handling of security research is collapsing responsible disclosure timelines and pushing vulnerabilities straight into the hands of attackers.

Stance · BearishConfidence · Established

The article concludes that Microsoft’s punitive approach to vulnerability reporting guarantees faster exploitation, degraded enterprise visibility, and a permanent fracture in the researcher-vendor relationship.

Key takeaways

  • Attackers can steal unscoped OAuth tokens and gain full read-write access to all linked GitHub repositories by injecting a malicious extension through github.dev’s Jupyter notebook viewer.
  • Standard patching windows have effectively vanished, reducing enterprise defense cycles from weeks to mere hours as coordinated disclosure breaks down.
  • Microsoft’s invocation of the Digital Crimes Unit against independent researchers has triggered severe industry backlash and threatens to permanently damage voluntary bug bounty participation.
  • The Nightmare Eclipse collective has warned of additional disruptive zero-day releases tied to unresolved grievances, testing the resilience of Microsoft’s security feedback loops.

What to watch next

  • Whether Microsoft modifies its disclosure policies before the July 14 deadline triggers further leaks
  • Adoption of stricter cryptographic signing and sandboxing requirements for VS Code extensions
  • Industry-wide migration toward formalized, legally protected safe harbors for independent researchers

Who should care

Enterprise security teamsDevOps and platform engineersVulnerability disclosure managers

Key players

Ammar AskarMicrosoft Security Response CenterNightmare EclipseVisual Studio CodeDigital Crimes Unit

Auto-generated from the article by our model — a reading aid, not a replacement for the piece.

The dispatch

One sharp read on the day’s biggest tech story.

Reported analysis for people who build software — free, most days, no spam.

Support our workIndependent, reader-funded tech journalism. If a piece helped you, chip in.Chip in →