PoC Monitor
← All CVEs
CVE-2026-21852published 21 Jan 2026 20:42 UTCData: JSON · API

Anthropics claude-code

Claude Code Leaks Data via Malicious Environment Configuration Before Trust Confirmation

Claude Code is an agentic coding tool. Prior to version 2.0.65, vulnerability in Claude Code's project-load flow allowed malicious repositories to exfiltrate data including Anthropic API keys before users confirmed trust. An attacker-controlled repository could include a settings file that sets ANTHROPIC_BASE_URL to an attacker-controlled endpoint and when the repository was opened, Claude Code would read the configuration and immediately issue API requests before showing the trust prompt, potentially leaking the user's API keys. Users on standard Claude Code auto-update have received this fix already. Users performing manual updates are advised to update to version 2.0.65, which contains a patch, or to the latest version.

First public PoC 36 daysafter publication · 27 Feb 2026
Public PoCs11 on GitHub · +2 suspicious, not counted
CVSS7.5high · scored by NVD · CWE-522
CISA KEV Not listednot known exploited

Affected versions and mitigation

From the CVE record and the vendor's references.

Affected

anthropics claude-code
  • < 2.0.65

Fixed in 2.0.65

Mitigation

Upgrade claude-code to 2.0.65 or later.

References

Timeline

Days relative to CVE publication
t=0+7+14+21+28+35CVE published21 Jan 2026 · t = 0First public PoC27 Feb 2026 · t+36d

First public PoC 36 days after publication; not in CISA KEV. First PoC: atiilla/CVE-2026-21852-PoC (created 27 Feb 2026, found by us: backfilled).

PoC repositories

Found on GitHub.

1 PoCs · 2 pending · 2 suspicious or fake

PoCs 1
RepositoryCreatedStarsAssessment
atiilla/CVE-2026-21852-PoC 27 Feb 2026 +36 d 24 PoC
Pending 2
RepositoryCreatedStarsAssessment
M0broot/CVE-Archive 27 Feb 2026 +37 d 0 Pending · repository unavailable (HTTP 404)
abhishek2512mishra/claude-code-security-audit 13 Sep 2026 +235 d 0 Pending
Flags · score 15
  • no network code found (normal for a local exploit)

Below the suspicion threshold (50); does not change the assessment.

Suspicious or fake 2
RepositoryCreatedStarsAssessment
TreRB/ai-ide-config-guard 20 Apr 2026 +88 d 0 Suspicious · score 100
Why suspicious
  • pipes a download into a shell / encoded PowerShell in README.md
  • pipes a download into a shell / encoded PowerShell in scripts/demo-vuln-scan.mjs
  • pipes a download into a shell / encoded PowerShell in src/checks/g7_shell_in_rules.js
  • pipes a download into a shell / encoded PowerShell in test/g1_claude_hooks.test.js
  • pipes a download into a shell / encoded PowerShell in test/g4_vscode_tasks.test.js
  • pipes a download into a shell / encoded PowerShell in test/g7_shell_in_rules.test.js
  • pipes a download into a shell / encoded PowerShell in test/integration.test.js
Perufitlife/dotclaude-security 21 Jun 2026 +151 d 0 Suspicious · score 100
Why suspicious
  • pipes a download into a shell / encoded PowerShell in scripts/audit.js
  • pipes a download into a shell / encoded PowerShell in test/audit.test.js

Only PoCs count toward the first-PoC date and the characteristics below. Suspicious repositories (score ≥ 50) never count. Do not run PoC code from unknown sources.