External risk intelligence

PraisonAI Claude GitHub Actions Workflow Command Injection

CVE advisorySeverity: CRITICAL (CVSS 10.0)

CVE-2026-48168

This vulnerability exists within a build-time GitHub Actions workflow. It is triggered by internal CI/CD processes rather than a public-facing service, web application, or edge gateway reachable from the internet.

Command Injection

Halo Surface Signal: 1 out of 5 — much less likely to be public-facing.

External exposure likelihood

Horizon Alert

Summary of the vulnerability and why it matters

A security vulnerability has been identified in Prais<ctrl62>.

  • Unvalidated branch names in workflows pose a risk.
  • It impacts code integrity and access controls.
  • Focus on confirming system relevance and exposure.

Attack Path

How an attacker could exploit the issue

An attacker can initiate a compromise by creating a malicious pull request with a specially crafted branch name. By including shell metacharacters in the branch name and commenting on the pull request, the attacker can trick the GitHub Actions runner into executing arbitrary commands. This allows for potential compromise of the repository, including writing to it, altering issues and pull requests, or abusing authentication tokens.

  • Requires external contributor access.
  • Triggered by a malicious pull request comment.
  • Risk includes code execution and token abuse.

Live Threat

Current exploitation, exposure, and threat context

This vulnerability could allow an attacker to execute arbitrary shell commands on GitHub Actions runners. These commands could then be used to compromise later privileged steps within the workflow, potentially leading to unauthorized modifications of the repository, manipulation of pull requests and issues, or abuse of OIDC tokens.

  • Repository data and access controls.
  • Malicious code injection via branch names.
  • Unauthorized repository changes or token abuse.

Operational Fix

Recommended remediation, mitigation, and detection steps

The PraisonAI system's GitHub Actions workflow is vulnerable to command injection, potentially allowing unauthorized code execution and repository manipulation. This impacts the security of the CI/CD pipeline and any code or secrets managed within the GitHub repository. The first practical step is to identify all instances of PraisonAI and determine if they are susceptible to this workflow exploitation, then prioritize remediation based on risk.

  • Platform or application owners should lead remediation.
  • Verify if affected PraisonAI versions are in use.
  • Coordinate vendor engagement for secure updates.

Supplementary metadata

Validate whether this threat affects your internet-facing exposure.

Halo Threat Intelligence helps prioritize remediation with Halo Surface Signal and H/A/L/O context. Start exposure validation with a free external attack surface trial.

Frequently asked questions

What is PraisonAI and how does it use GitHub Actions?

PraisonAI is a framework for managing multi-agent AI teams. It helps automate complex tasks by coordinating various AI agents. It uses GitHub Actions as part of its development lifecycle, specifically for automating workflows like interacting with Claude to assist in repository maintenance and pull request management.

What is the vulnerability in CVE-2026-48168?

This CVE involves a command injection vulnerability, formally categorized as CWE-862 (Missing Authorization). The system fails to properly validate user-supplied input—specifically GitHub pull request branch names—before processing them in a Bash script. This allows malicious input to be executed directly by the system runner.

How does an attacker trigger this command injection?

An attacker triggers the bug by creating a pull request from a fork with a branch name containing specific shell metacharacters and adding a @claude comment. It does not trigger if a user simply views the repository or if the pull request branch name contains only standard, non-malicious characters.

Is this CVE-2026-48168 a risk to my internet-facing services?

According to Halo Surface Signal, this vulnerability is considered very unlikely to impact internet-facing services. Because the flaw resides in a build-time CI/CD workflow rather than a web application or edge gateway, the attack surface is generally limited to the repository's internal automated testing and integration processes.

How do I secure my environment against this issue?

The primary step is to upgrade your PraisonAI installation to version 4.6.40 or newer, which contains the necessary workflow fixes. You should inventory your repositories to identify where these automated workflows are active and verify that the updated, secure code has been successfully integrated into your CI/CD pipelines.

References