Skip to main content
Mallory
HighVulnerability

CVE-2026-55607: Claude Code sandbox escape via Git worktree path confusion

Published Jun 29, 2026 · Updated 1 month ago
Trending this weekNot trendingOutside this week’s top 100
Mention activity · 30 daysNo mention activity in the last 30 days.
CVSS7.7High · v4
EPSS0.69%53rd percentile
CISA KEVNoNot listed
In the wildNoneNo exploitation reported
Vendor
Anthropic
IdentifiersCVE-2026-55607CWE-22 · Improper Limitation of a Pathname…

Claude Code versions 2.1.38 through versions before 2.1.163 contain a Git worktree path-confusion vulnerability. Worktree handling permits names corresponding to Git metadata directories and navigation to worktrees outside the sandbox context. An attacker can combine this behavior with symlink manipulation and Git fsmonitor execution during worktree operations to overwrite files in the user's home directory and execute code outside Seatbelt sandbox restrictions. Reliable exploitation requires a user to clone a malicious repository containing prompt-injection content and run Claude Code against it.

Do you run this?See affected versions in your environment and whether adversaries are exploiting this vulnerability against assets like yours.
Check your exposure
Analyst brief

Impact, mitigation & remediation

What it means and what to do now: patch path, interim mitigations, and the assume-compromise checklist.

Impact

What an attacker gets, and what they’ve been doing with it.

Successful exploitation breaks the Claude Code sandbox boundary, allowing attacker-controlled writes to files in the user's home directory and code execution outside Seatbelt restrictions. Modification of shell initialization files can cause attacker-controlled code to execute when the affected shell initializes, potentially providing persistence with the user's privileges.

Mitigation

If you can’t patch tonight, do this now.

Until the update is installed, avoid cloning untrusted repositories and running Claude Code against them, particularly repositories containing prompt-injection content.

Remediation

Patch, then assume compromise.

Upgrade the @anthropic-ai/claude-code package to version 2.1.163 or later. Standard Claude Code auto-update delivered the fix automatically; verify that the installed version includes it.

Public exploits

Exploits

No public exploits tracked yet. Mallory keeps watching.

View more in app →
No public exploit code observed for this vulnerability. Mallory keeps watching.
Exposure surface

Affected products & vendors

Products and vendors Mallory has correlated with this vulnerability. Open in Mallory to drill down to specific CPE configurations and version ranges.

VendorProductType
AnthropicClaude Codeapplication
FAQ

Frequently asked questions

Is CVE-2026-55607 exploited in the wild?
As of Oct 16, 2026, Mallory has no reports of CVE-2026-55607 being exploited in the wild. It is not listed in CISA KEV.
Is there a public exploit for CVE-2026-55607?
As of Oct 16, 2026, Mallory has found no public exploit for CVE-2026-55607.
Is there a patch for CVE-2026-55607?
Upgrade the @anthropic-ai/claude-code package to version 2.1.163 or later. Standard Claude Code auto-update delivered the fix automatically; verify that the installed version includes it.
What products are affected by CVE-2026-55607?
Anthropic Claude Code.
How severe is CVE-2026-55607?
CVE-2026-55607 has a CVSS v4 score of 7.7 (High) and an EPSS score of 0.69% (53rd percentile) as of Oct 16, 2026.
Activity feed

Recent activity

11 sources tracked across advisories, community write-ups, and news. New activity surfaces here as Mallory finds it.

View more in app →
Do you run this?See affected versions in your environment and whether adversaries are exploiting this vulnerability against assets like yours.
Check your exposure
Mallory platform

The version that knows your environment.

This page is what’s public. Mallory adds the parts that aren’t: which of your assets are affected, which adversaries are exploiting it right now, and which detections to deploy tonight.
Exposure mappingQuery your assets running an affected version, and investigate the blast radius.
Threat actor evidenceEvery report linking this CVE to a named adversary.
Associated malwareMalware families linked to this CVE, with evidence and IOCs.
Detection signaturesYARA, Sigma, Snort, and vendor rules, auto-deployed to your SIEM.
Vendor-by-vendor mappingCross-references every affected SKU, including bundled OEM variants.
Social activity10Community discussion across Reddit, Mastodon, and other social sources.