/

CVE-2026-32981

Severity:

HIGH RISK

(7.5)

Path Traversal in Ray Dashboard Static File Handler

CVE-2026-32981

Modified

Vulnerability type

Category:

Path Traversal

CWE:

CWE-22

Impact:

Local File Disclosure

Severity:

HIGH RISK

(7.5)

CVSS version:

3.1

Catch this CVE while code is being written

Detect vulnerable code directly in pull requests, so CVEs are fixed before merge.

Summary

A path traversal vulnerability in the Ray Dashboard's static file handling allows user-controlled paths to escape the intended web root directory.

By supplying traversal sequences such as "../", an attacker can potentially read arbitrary files accessible to the Ray Dashboard process, leading to sensitive local file disclosure.

Affected context

This vulnerability occurs in Ray deployments exposing the Ray Dashboard (default port 8265), where the server serves static resources from a designated directory.

When user-supplied path segments are incorporated into filesystem lookups without robust normalization and validation, requests can reference files outside the static root.

The risk is higher on hosts where Ray runs with access to sensitive configuration files, credentials, logs, or other application data.

Exploit preconditions

  • Attack surface: HTTP interface of the Ray Dashboard (typically reachable on port 8265).
  • Required access: Ability to send crafted HTTP requests to the dashboard endpoint; no special privileges beyond network reachability.
  • Trigger point: Static file or asset retrieval endpoints that accept path parameters or query values used to construct filesystem paths.
  • Why exploitation is possible: The server accepts user-controlled paths and concatenates them with a base directory without enforcing strict canonicalization or blocking traversal patterns, allowing directory escapes.

Why this happens

The underlying defect stems from constructing filesystem paths directly from user input and the static root directory without proper normalization and bounds checking.

User-controlled strings containing sequences like "../" or URL-encoded equivalents are resolved by the operating system's path handling to directories above the intended static folder.

This behavior aligns with CWE-22: Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal'), where an application fails to constrain file access to a safe subset of the filesystem.

Potential impact

  • Disclosure of application configuration files that may contain secrets or connection strings.
  • Access to environment-specific files such as logs, internal diagnostics, or system information.
  • Exposure of credentials, tokens, or SSH keys if stored in accessible directories.
  • Assistance in further attacks by revealing filesystem layout, user names, or other reconnaissance data.

General mitigation guidance

Effective mitigation focuses on constraining file access to a well-defined static directory and rejecting unsafe paths:

  • Normalize and canonicalize requested paths, then verify they remain within the configured static root before access.
  • Reject any path containing traversal patterns or ambiguous segments (e.g., "..", ".", repeated slashes), including their encoded representations.
  • Use framework-provided safe file-serving utilities that enforce a document root boundary instead of building paths manually.
  • Limit the Ray Dashboard process privileges and filesystem visibility, ensuring it cannot read highly sensitive system or credential files.
  • Store secrets and critical configuration outside of directories that web-facing components can reach.

How teams detect issues like this

Organizations can identify and monitor for path traversal issues through a mix of static, dynamic, and operational checks:

  • Static analysis rules that flag direct concatenation of user input into filesystem APIs without normalization or boundary checks.
  • Code review checklists that require explicit verification of file-serving logic, ensuring the use of canonicalization and root-directory enforcement.
  • Dynamic security testing or fuzzing that probes static file endpoints with traversal patterns (e.g., "../", encoded variants) and observes responses.
  • Web server and application logging tuned to highlight suspicious path requests containing traversal indicators or unusual directory patterns.

Sources

SourceLink
github.comgithub.com/ray-project/ray
packetstorm.newspacketstorm.news/files/id/215801/
vulncheck.comvulncheck.com/advisories/ray-dashboard-path-traversal-leading-to-local-file-disclosure
access.redhat.comaccess.redhat.com/errata/RHSA-2026:19712
access.redhat.comaccess.redhat.com/errata/RHSA-2026:24977
access.redhat.comaccess.redhat.com/errata/RHSA-2026:42644
access.redhat.comaccess.redhat.com/errata/RHSA-2026:5809
access.redhat.comaccess.redhat.com/errata/RHSA-2026:6761
access.redhat.comaccess.redhat.com/errata/RHSA-2026:6762
access.redhat.comaccess.redhat.com/security/cve/CVE-2026-32981
bugzilla.redhat.combugzilla.redhat.com/show_bug.cgi?id=2448440
security.access.redhat.comsecurity.access.redhat.com/data/csaf/v2/vex/2026/cve-2026-32981.json

Frequently asked

questions

Yes. Any attacker who can reach the Ray Dashboard over the network and send HTTP requests can attempt to exploit the path traversal behavior.

The attacker only needs network access to the dashboard port (typically 8265). No authenticated dashboard access is strictly required if the vulnerable endpoints are exposed anonymously.

Enforce strict path canonicalization and verification against a static root directory, and rely on hardened file-serving helpers instead of manual path concatenation. Any request that resolves outside the configured static directory should be rejected.

Treat all path components from users as untrusted, adopt secure-by-default abstractions for file access, incorporate static analysis rules for path-based APIs, and include path traversal tests in regular security and regression testing.

Find out what's already
exploitable in your codebase.