VYPR
Medium severity5.0NVD Advisory· Published May 17, 2026· Updated May 19, 2026

CVE-2026-8767

CVE-2026-8767

Description

A vulnerability has been found in vercel ai up to 3.0.97. Impacted is the function run of the file .github/workflows/prettier-on-automerge.yml of the component PR Branch Name Interpolation. The manipulation leads to os command injection. The attack can be initiated remotely. The complexity of an attack is rather high. The exploitability is considered difficult. The exploit has been disclosed to the public and may be used. The vendor was contacted early about this disclosure but did not respond in any way.

AI Insight

LLM-synthesized narrative grounded in this CVE's description and references.

A command injection vulnerability in vercel ai's GitHub Actions workflow allows attackers to execute arbitrary commands via malicious PR branch names.

Vulnerability

The vulnerability resides in the GitHub Actions workflow file .github/workflows/prettier-on-automerge.yml of vercel ai up to version 3.0.97. The run step uses unsafe direct string interpolation of the user-controlled branch name (github.event.pull_request.head.ref) via ${{ }} syntax, which is evaluated before the shell script executes. This allows an attacker to inject arbitrary OS commands by crafting a branch name containing subshell sequences such as $() combined with $IFS to bypass branch naming restrictions [1].

Exploitation

An attacker must be able to submit a pull request to the repository. The attack is remote but has high complexity due to branch name constraints (no spaces or quotes). However, the attacker can embed command substitution using $() and $IFS to achieve code execution. When the workflow triggers on a pull request event, the malicious branch name is interpolated into the bash script, and the injected commands execute in the runner sandbox [1].

Impact

Successful exploitation allows the attacker to execute arbitrary bash commands within the GitHub Actions runner sandbox. This can lead to unauthorized modification of repository contents, compromise of secrets or credentials accessible to the runner, and other actions depending on the runner's permissions [1].

Mitigation

No official fix has been released by the vendor (vercel), who did not respond to the disclosure. The recommended mitigation is to pass all dynamic attacker-controllable input via intermediate environment variables (using the env: contextual map) instead of direct string interpolation. Users should update to a version beyond 3.0.97 if available, or manually modify the workflow file to apply the remediation [1].

AI Insight generated on May 21, 2026. Synthesized from this CVE's description and the cited reference URLs; citations are validated against the source bundle.

Affected products

1

Patches

0

No patches discovered yet.

Vulnerability mechanics

AI mechanics synthesis has not run for this CVE yet.

References

4

News mentions

0

No linked articles in our index yet.