CVE-2026-82862 Overview
CVE-2026-82862 affects Hulumi versions before v1.3.2. The vulnerability stems from unsafe resolution of the threat-model helper script from an untrusted root directory. Attackers can place malicious files inside the workspace to shadow the intended helper script. When a user runs local skill execution, Hulumi loads the attacker-controlled file instead of the legitimate helper, resulting in arbitrary code execution in the user's local context. The flaw is tracked under CWE-426: Untrusted Search Path.
Critical Impact
A local attacker who can write files into a Hulumi workspace can execute arbitrary code during local skill execution, compromising confidentiality, integrity, and availability of the affected host.
Affected Products
- Hulumi versions prior to v1.3.2
- Local skill execution workflows relying on the threat-model helper script
- Workspaces where untrusted files can be introduced through shared repositories or collaborators
Discovery Timeline
- 2026-08-31 - CVE-2026-82862 published to NVD
- 2026-09-01 - Last updated in NVD database
Technical Details for CVE-2026-82862
Vulnerability Analysis
Hulumi's local skill execution flow invokes a threat-model helper script. Prior to v1.3.2, Hulumi resolves this helper from an unsafe root, meaning workspace-local files are searched before or in place of the trusted helper location. An attacker who can add or modify files within the workspace can plant a file with the expected helper name. When the user triggers local skill execution, Hulumi loads and runs the attacker's file instead of the intended helper. Because execution happens under the local user account, the attacker inherits that user's privileges, file access, and network reach.
Root Cause
The root cause is an untrusted search path issue classified as CWE-426. Hulumi resolves the helper script using a lookup rooted in the workspace rather than a fixed, trusted installation path. Any file placed in a resolvable workspace location can shadow the intended helper. The application does not validate the origin, integrity, or expected path of the helper before executing it.
Attack Vector
Exploitation requires local access to a workspace that Hulumi will operate on. Attack scenarios include a malicious commit added to a shared repository, a poisoned project template, or a file dropped into a shared development directory. The attacker needs no authentication to Hulumi itself and no user interaction beyond the victim invoking local skill execution against the poisoned workspace. Successful exploitation yields arbitrary code execution with the victim's privileges.
See the GitHub Security Advisory GHSA-mjcg-x5mr-27ww and the VulnCheck Advisory on Hulumi for additional technical context. No public proof-of-concept exploit code is available at the time of publication.
Stop Remote Code Execution Before It Runs
Singularity™ Endpoint detects and blocks exploit attempts at machine speed, including zero-days and novel attack techniques.
Stop Exploit AttemptsDetection Methods for CVE-2026-82862
Indicators of Compromise
- Unexpected script files matching the threat-model helper name inside project workspaces or repositories
- New or modified files in workspace roots introduced through untrusted commits, pull requests, or shared templates
- Child processes spawned by the Hulumi binary that do not match the expected helper's cryptographic hash or installation path
Detection Strategies
- Compare the resolved helper script path at runtime against the trusted Hulumi installation directory and alert on mismatches
- Monitor endpoint telemetry for Hulumi invoking interpreters (for example python, node, or shell) against files located under user-writable workspace paths
- Audit source control for additions of files whose names collide with Hulumi helper scripts
Monitoring Recommendations
- Enable process-lineage logging on developer workstations to trace helper script execution back to the Hulumi parent process
- Alert on writes to workspace directories that create files with helper-script naming patterns immediately before a skill execution event
- Track installed Hulumi versions across the estate and flag hosts running versions earlier than v1.3.2
How to Mitigate CVE-2026-82862
Immediate Actions Required
- Upgrade Hulumi to v1.3.2 or later on every host that runs local skills
- Inventory shared workspaces and repositories for files that shadow the threat-model helper script and remove any that were not authored by trusted maintainers
- Restrict write access to workspace directories to trusted users and processes until patching is complete
Patch Information
Hulumi v1.3.2 fixes the unsafe helper resolution. Refer to the GitHub Security Advisory GHSA-mjcg-x5mr-27ww for release details and upgrade guidance.
Workarounds
- Avoid running local skill execution against workspaces sourced from untrusted repositories, forks, or contributors
- Review incoming commits and pull requests for files that match helper script names before running Hulumi against the branch
- Execute Hulumi inside an isolated sandbox or ephemeral container so that helper shadowing cannot affect the host filesystem or credentials
Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
