Loading vulnerability…
CVE-2025-53773 is a command injection vulnerability affecting GitHub Copilot and Visual Studio. The flaw stems from indirect prompt injection and unsafe handling of untrusted repository content, allowing attacker-controlled instructions embedded in developer-consumed content to influence the assistant’s behavior. Reported exploitation paths include malicious repository rules or hidden instructions in pull request descriptions, README content, or similar project artifacts that cause the assistant to modify IDE configuration and enable automatic approval behavior such as Copilot's experimental auto-approve or "YOLO" mode. Once confirmations are suppressed, the assistant can be induced to invoke shell commands locally, resulting in command execution on the developer workstation. The issue has been described as enabling local arbitrary code execution across major desktop platforms when a developer opens or processes attacker-controlled project content with the vulnerable assistant integration.
Mallory correlates every CVE against your assets, your vendors, and active adversary campaigns. Know which vulnerabilities matter for you, not just which ones are loud.
What it means. What to do now. Patch path, mitigations, and the assume-compromise checklist.
What an attacker gets, and what they’ve been doing with it.
If you can’t patch tonight, do this now.
Patch, then assume compromise.
No valid public exploits. Mallory filtered out 1 candidate as fakes, detection scripts, or README-only repos.
All candidate exploits were filtered out by Mallory's validation.
Products and vendors Mallory has correlated with this vulnerability. Open in Mallory to drill down to specific CPE configurations and version ranges.
Vendor-confirmed product mapping. Mallory continuously reconciles this list against your asset inventory.
62 sources tracked across advisories, community write-ups, and news. New activity surfaces here as Mallory finds it.
A command injection vulnerability affecting GitHub Copilot / Visual Studio workflow, where a malicious rules file in a repository can cause local arbitrary code execution when a developer opens the project and Copilot processes it.
A cited high-severity vulnerability associated with indirect prompt injection in AI agents, where the agent trusted untrusted content.
A separate previously patched GitHub Copilot vulnerability that allowed prompt injection to enable YOLO mode by modifying workspace settings.
A GitHub Copilot vulnerability where hidden instructions in pull request descriptions could manipulate Copilot into changing settings to auto-approve actions, disabling confirmations and enabling unrestricted shell execution.
Query your assets running an affected version, and investigate the blast radius.
Every observed campaign linking this CVE to a named adversary.
Malware families riding this exploit, with evidence and IOCs.
YARA, Sigma, Snort, and vendor rules, auto-deployed to your SIEM.
Cross-references every affected SKU, including bundled OEM variants.
Community discussion across Reddit, Mastodon, and other social sources.