CVE-2026-57495

AgenticMail: Unauthenticated inbound mail triggers bypassPermissions resume of the operator's Claude Code session (bridge-wake)

Published Jul 20, 2026 CVSS 4.0 · 8.2 High EPSS < 1% KEV · Not listed
Description
AgenticMail gives AI agents real email addresses and phone numbers. In @agenticmail/claudecode prior to version 0.2.39, @agenticmail/codex prior to version 0.1.33, @agenticmail/core prior to version 0.9.43, and @agenticmail/openclaw prior to version 0.5.71, two inbound-mail handlers act on a privileged effect without verifying that the sender is the operator, while a sibling handler in the same repo does. The higher-impact one: any external email routed to the bridge inbox causes the dispatcher to resume the operator's Claude Code session with `permissionMode: 'bypassPermissions'`, embedding the attacker-controlled `from`/`subject`/`preview` verbatim into the prompt the resumed agent reads — an indirect prompt injection into a fully-privileged agent (Bash/Write/Edit/WebFetch + the agenticmail MCP toolbelt) running as the operator's OAuth identity. The sibling operator-query email-reply hook gates the same untrusted-From provenance with `isOperatorReplySender(replyFrom, config.operatorEmail)` (fail-closed); the bridge-wake path — a strictly higher-privilege effect — has no equivalent. @agenticmail/claudecode 0.2.39, @agenticmail/codex 0.1.33, @agenticmail/core 0.9.43, and @agenticmail/openclaw 0.5.71 contain a fix.

Analysis

AI-generated

Technical impact

AgenticMail's @agenticmail/claudecode, @agenticmail/codex, @agenticmail/core, and @agenticmail/openclaw packages before specific versions contain a flaw where inbound mail handlers execute privileged operations without verifying the sender. This allows any external email to trigger a dispatcher that resumes the operator's Claude Code session with a bypassPermissions mode, embedding attacker‑provided fields directly into the prompt the agent consumes. The attacker can thus inject malicious instructions into a fully‑privileged agent running as the operator’s OAuth identity, potentially enabling remote code or system execution. The weakness corresponds to CWE‑306, reflecting lack of authentication.

Affected Systems

Vendors: AgenticMail. Products affected include the @agenticmail/claudecode library up to version 0.2.38, @agenticmail/codex up to version 0.1.32, @agenticmail/core up to 0.9.42, and @agenticmail/openclaw up to 0.5.70. These libraries are used to provide AI agents with real email addresses and phone numbers. Systems using any of these packages below the specified versions are vulnerable until patched to the fixed releases.

Risk and Exploitability

With a CVSS score of 8.2 the flaw rates as high severity, and an EPSS score of less than 1% suggests that exploitation is currently unlikely but not impossible. Because the vulnerability relies on unauthenticated inbound email to trigger privileged behavior, an attacker with the ability to send mail to the bridge inbox can initiate the bypass with little-to-no preconditions, effectively giving them a path to inject arbitrary text into a privileged agent session. Although it is not listed in CISA's KEV catalog, the potential for remote code execution and the high privilege level of the affected agent warrant careful attention. Upgrading to the patched releases eliminates the risk.

Remediation

No solution or workaround provided in the CVE record.

OpenCVE Recommended Actions

  • Apply the published patch by upgrading @agenticmail/claudecode to version 0.2.39 or later.
  • Upgrade @agenticmail/codex to version 0.1.33 or later.
  • Update @agenticmail/core to version 0.9.43 or later and @agenticmail/openclaw to version 0.5.71 or later.
  • Ensure inbound email processing enforces sender authentication to avoid future bypasses.

Generated by OpenCVE AI on July 30, 2026 at 18:20 UTC.

Tracking

Sign in to view the affected projects.

Advisories

Source ID Title
Github GHSA Github GHSA GHSA-fq4x-789w-jg5h AgenticMail: Unauthenticated inbound mail triggers bypassPermissions resume of the operator's Claude Code session (bridge-wake)
History
Type Values Removed Values Added
First Time appeared Agenticmail
Agenticmail agenticmail
Agenticmail claudecode
Agenticmail codex
Agenticmail openclaw
Vendors & Products Agenticmail
Agenticmail agenticmail
Agenticmail claudecode
Agenticmail codex
Agenticmail openclaw
Type Values Removed Values Added
Metrics ssvc

{'options': {'Automatable': 'yes', 'Exploitation': 'none', 'Technical Impact': 'total'}, 'version': '2.0.3'}

Type Values Removed Values Added
Description AgenticMail gives AI agents real email addresses and phone numbers. In @agenticmail/claudecode prior to version 0.2.39, @agenticmail/codex prior to version 0.1.33, @agenticmail/core prior to version 0.9.43, and @agenticmail/openclaw prior to version 0.5.71, two inbound-mail handlers act on a privileged effect without verifying that the sender is the operator, while a sibling handler in the same repo does. The higher-impact one: any external email routed to the bridge inbox causes the dispatcher to resume the operator's Claude Code session with `permissionMode: 'bypassPermissions'`, embedding the attacker-controlled `from`/`subject`/`preview` verbatim into the prompt the resumed agent reads — an indirect prompt injection into a fully-privileged agent (Bash/Write/Edit/WebFetch + the agenticmail MCP toolbelt) running as the operator's OAuth identity. The sibling operator-query email-reply hook gates the same untrusted-From provenance with `isOperatorReplySender(replyFrom, config.operatorEmail)` (fail-closed); the bridge-wake path — a strictly higher-privilege effect — has no equivalent. @agenticmail/claudecode 0.2.39, @agenticmail/codex 0.1.33, @agenticmail/core 0.9.43, and @agenticmail/openclaw 0.5.71 contain a fix.
Title AgenticMail: Unauthenticated inbound mail triggers bypassPermissions resume of the operator's Claude Code session (bridge-wake)
Weaknesses CWE-306
References
Metrics cvssV4_0

{'score': 8.2, 'vector': 'CVSS:4.0/AV:N/AC:H/AT:N/PR:N/UI:N/VC:N/VI:H/VA:N/SC:N/SI:N/SA:N'}

Weaknesses
  • CWE-306

    Missing Authentication for Critical Function

Sources

Status
PUBLISHED
Assigner
GitHub_M
Published
Updated
Jul 22, 2026, 1:53 PM
Reserved
Jun 24, 2026, 2:53 PM
Status
Deferred
Published
Jul 20, 2026, 10:17 PM
Modified
Jul 23, 2026, 3:54 PM