harness-opencode 1.0.27 JS Sandbox Remote Commands via Path Auth Fallback
CVE-2026-64651 Published on July 20, 2026

AI SDK OpenCode Harness Tool Relay Authorization Bypass
The `@ai-sdk/harness-opencode` tool connects HarnessAgent to OpenCode through a sandboxed bridge. Prior to version 1.0.28, the tool relay authorizes requests from any process whose command line contains an allowed helper script path (`host-tool-mcp.mjs`). This allows untrusted code executing in the sandbox to invoke arbitrary host-exposed tools including secret lookups, deployment operations, and cloud API calls without a corresponding model-authorized tool-call event. Exploitation requires a Linux environment (the vulnerable fallback checks `process.platform === 'linux'` and reads `/proc`); an active harness session with one or more host-provided tools; and untrusted code executing in the sandbox (e.g. a malicious dependency, build script, or lifecycle hook) The fix in version 1.0.28 removes the process-path authorization fallback entirely. Relay requests are now only accepted after exact, short-lived, one-time authorization matching the tool name and input from a bridge-observed model event. Some workarounds are available. Do not run the OpenCode harness on untrusted repositories or with untrusted dependencies, and/or limit host-exposed tools to non-sensitive operations when working with untrusted code.

Vendor Advisory NVD

Vulnerability Analysis

CVE-2026-64651 can be exploited with local system access, and does not require authorization privileges or user interaction. This vulnerability is considered to have a low attack complexity. The potential impact of an exploit of this vulnerability is considered to have no impact on confidentiality and integrity and availability.

Exploitation:
NONE
Automatable:
NO
Technical Impact:
PARTIAL
Attack Vector:
LOCAL
Attack Complexity:
LOW
Privileges Required:
NONE
User Interaction:
NONE

Weakness Type

What is an AuthZ Vulnerability?

The software performs an authorization check when an actor attempts to access a resource or perform an action, but it does not correctly perform the check. This allows attackers to bypass intended access restrictions.

CVE-2026-64651 has been classified to as an AuthZ vulnerability or weakness.


Products Associated with CVE-2026-64651

Want to know whenever a new CVE is published for Vercel Ai? stack.watch will email you.

Affected Versions

vercel @ai-sdk/harness-opencode Version < 1.0.29 is affected by CVE-2026-64651

Exploit Probability

EPSS
0.17%
Percentile
6.93%

EPSS (Exploit Prediction Scoring System) scores estimate the probability that a vulnerability will be exploited in the wild within the next 30 days. The percentile shows you how this score compares to all other vulnerabilities.