CVE-2026-105037: EnergizeLab Station-Middleware Arbitrary File Read Exposes Server Secrets

Executive Summary

A critical arbitrary file-read vulnerability has been disclosed in EnergizeLab station-middleware, the NestJS-based Docker service that backs EnergizeLab's cloud infrastructure for its Eilik and EilikO companion-robot line (the "AI Station" charging/docking hardware and its mobile/desktop companion apps talk to this middleware). Tracked as CVE-2026-105037, the flaw lets any registered, low-privilege account — not just administrators — supply an attacker-controlled file path and read arbitrary files off the server, exposing middleware configuration files and server secrets.

CVSS Score: 9.9 (Critical) — CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:H

The vulnerability was published to NVD on 2026-10-15 as part of a broader disclosure covering the EnergizeLab Eilik/EilikO platform, researched between June and September 2026. At the time of writing there is no confirmed patched release, no CISA KEV listing, and no public proof-of-concept exploit — researchers point to a dedicated GitHub repository (see Sources) for further technical detail rather than a step-by-step writeup in the NVD record itself.

Vulnerability Overview

Affected Versions and Patches

The confirmed vulnerable range is station-middleware 1.6.5 ≤ v ≤ 1.8.2 (Docker image). As of this writing, research sources we could find do not confirm a fixed version number. Multiple trackers (VulDB, independent CVE-intelligence writeups) explicitly note the absence of a verified patch release at publication time, with guidance pointing organizations toward interim mitigations (restrict access, segment the service) "until a verified patch is deployed." We did not find a GitHub Security Advisory (GHSA) ID specifically minted for CVE-2026-105037 — a sibling CVE in the same disclosure batch (CVE-2026-105038) does have one (GHSA-33f8-3xw6-r37g), but this particular flaw appears to be tracked only through the NVD/CVE record and the researcher's own writeup.

If you run a self-hosted or vendor-managed instance of station-middleware, treat any version in the 1.6.5-1.8.2 range as vulnerable and check directly with EnergizeLab for a current patched build rather than assuming one exists.

How the Vulnerability Works

Per the NVD description and corroborating secondary analysis, the flaw sits in how station-middleware handles file-path input on at least one endpoint reachable by authenticated users:

1. Any registered account can reach the flaw. Unlike a typical admin-only file-management feature, this endpoint is accessible to low-privilege accounts — the CVSS vector's PR:L (Privileges Required: Low) confirms no elevated role is needed, only a standard login.
2. The service accepts an attacker-controlled file path. The application takes user-supplied input and uses it to construct a file path for a read operation, without sufficiently validating or sanitizing it.
3. Insufficient path sanitization allows escape from the intended directory. One independent technical analysis describes this as effectively combining an insecure direct object reference with directory-traversal sequences (e.g., ../../../ style paths) to resolve outside the restricted directory the application intended to confine reads to.
4. The read is not limited to "safe" files. Because the path is attacker-controlled, the accessible file set is not limited to user-owned data — it can reach middleware configuration files and secrets stored on the host or in the container.

We could not find the exact vulnerable route, parameter name, or a published request/response proof-of-concept in the sources available at the time of writing. The researchers' detailed technical writeup and any PoC material live in a separate GitHub repository (see Sources); the public advisory text itself focuses on impact rather than exploit mechanics.

Important distinction: this is a low-privilege-to-secrets escalation, not an unauthenticated one. An attacker needs a valid account on the platform — but given that station-middleware supports a consumer robotics product, "registered account" could mean any customer who signed up through the companion app, which is a much lower bar than it sounds.

Impact Assessment

Successful exploitation exposes:

• Middleware configuration and source files — potentially revealing internal architecture, file paths, and logic useful for further attacks.
• Server secrets — the NVD description explicitly calls out secret exposure; depending on what's reachable, this could include API keys, database credentials, or signing/session secrets used by the middleware.

This flaw does not stand alone. It was disclosed alongside CVE-2026-105039, a missing-RBAC vulnerability in the same station-middleware product (versions 1.6.5-1.8.2) that lets any authenticated low-privilege user hit administrative APIs directly and retrieve AI provider keys, and modify configuration and object-store contents. Chained together, these two flaws mean a low-privilege account could potentially read secrets via CVE-2026-105037 and use administrative API access via CVE-2026-105039 to directly alter cloud configuration — a combination that research summaries describe as enabling a "cloud-to-fleet" style escalation across EnergizeLab's backend.

Because station-middleware is cloud/backend infrastructure for a consumer IoT product line (robot companions and their charging "stations"), the practical blast radius extends beyond a single customer's account — exposed server secrets could affect the shared backend serving the entire fleet of devices, not just the attacker's own hardware.

No evidence of active, in-the-wild exploitation was found in any source reviewed, and the vulnerability is not currently listed in the CISA Known Exploited Vulnerabilities (KEV) catalog.

Recommendations

1. Patch immediately if a fixed build becomes available. Contact EnergizeLab directly to confirm current patch status for station-middleware — public sources did not confirm a fixed version as of this advisory.
2. Restrict new account registration on any internet-facing station-middleware instance until patched; this vulnerability requires only a low-privilege registered account, so limiting who can register meaningfully reduces exposure.
3. Audit file-serving and file-read endpoints in the middleware for path/parameter validation — specifically look for any endpoint that accepts a file path, filename, or similar input from an authenticated-but-unprivileged caller.
4. Rotate any secrets potentially exposed — API keys, credentials, or tokens stored in middleware configuration or environment files reachable by the service account should be treated as compromised and rotated.
5. Review account activity for unusual read patterns from low-privilege accounts, especially repeated requests with path-like parameters or traversal sequences (../).
6. Treat this together with CVE-2026-105039 — patching one without addressing the other leaves the "secrets exposure + admin API access" chain partially open on the same product.
7. Segment or restrict network access to the middleware service as an interim mitigation if patching is not immediately possible.

Key Takeaways

1. CVE-2026-105037 (CVSS 9.9) lets any registered, low-privilege account — not just admins — read arbitrary files from EnergizeLab station-middleware, exposing server secrets.
2. The affected range is confirmed as station-middleware 1.6.5 through 1.8.2; no confirmed patched version was found in public sources at the time of writing.
3. The root cause is classified as CWE-73 (External Control of File Name or Path) by NVD-aligned trackers — effectively a path-traversal-style flaw in how the service resolves user-supplied file paths.
4. This flaw was disclosed alongside CVE-2026-105039 (missing RBAC on the same product's admin APIs) — the two can chain into a broader cloud-backend compromise.
5. No public exploit code and no CISA KEV listing exist as of publication; treat this as urgent-but-not-yet-weaponized.
6. Organizations running station-middleware should prioritize vendor contact for patch status, restrict registration, and rotate any potentially exposed secrets now rather than waiting for confirmed in-the-wild exploitation.

Related Vulnerabilities

This CVE is one of several disclosed the same day affecting EnergizeLab's Eilik/EilikO platform:

• CVE-2026-105039 — Missing RBAC on the same station-middleware admin APIs, exposing AI provider keys (CVSS 9.6). See /security/energizelab-station-middleware-missing-rbac-cve-2026-105039.
• CVE-2026-105038 — Unauthenticated AI model administration on the Eilik/EilikO FastAPI inference service (CVSS 9.8). See /security/energizelab-eilik-fastapi-unauthenticated-model-admin-cve-2026-105038.
• CVE-2026-105032 — Unauthenticated path-traversal file write on the same FastAPI inference service (CVSS 9.1). See /security/energizelab-eilik-fastapi-path-traversal-file-write-cve-2026-105032.

(Links above are best-effort guesses at sibling article slugs — if a link 404s, search the Labs security archive for the CVE number directly.)

Sources

• NVD — CVE-2026-105037
• Strix.ai — CVE-2026-105037
• OffSeq Radar — CVE-2026-105037
• VulDB — CVE-2026-105037
• takeonme.org — Multiple EnergizeLab Eilik/EilikO Platform Vulnerabilities
• TheHackerWire — CVE-2026-105037 analysis
• GitHub — xn0tsa/your-secrets-are-safe-with-me (researcher repository)