Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
Sources: National Vulnerability Database, CISA KEV, Exploit-DB
CVE-2026-108372 Overview
CVE-2026-108372 is a cross-origin information disclosure vulnerability in LoopX, an open and provider-neutral lightweight state kernel and local-first control plane for loop engineering. The flaw affects loopx serve-status in loopx/status_server.py, which emits an Access-Control-Allow-Origin: * header on unauthenticated responses served from predictable loopback ports. A remote attacker who lures an operator to attacker-controlled web content can issue cross-origin reads against the local status server. The vulnerability is tracked under [CWE-346] (Origin Validation Error) and affects versions 0.0.1 through 0.4.4.
Critical Impact
Attackers can read control-plane state, goal and run metadata, local absolute paths, private Markdown files under registered goal roots, and dry-run configuration projections from a victim's browser.
Affected Products
- LoopX versions 0.0.1 through 0.4.4
- The loopx serve-status component (loopx/status_server.py)
- StatusRequestHandler._send_json() and do_OPTIONS() response paths
Discovery Timeline
- 2026-10-16 - CVE-2026-108372 published to NVD
- 2026-10-21 - Last updated in NVD database
Technical Details for CVE-2026-108372
Vulnerability Analysis
The LoopX status server listens on a predictable loopback port and returns JSON payloads to local clients. The handler attaches Access-Control-Allow-Origin: * to every unauthenticated response, including preflight OPTIONS replies. Any website a victim visits can therefore use fetch() or XMLHttpRequest to issue cross-origin reads against http://127.0.0.1:<port>/ and parse the response body.
The exposed routes include GET /status.json, GET /review-material?goal_id=...&path=..., GET /, and POST /control-plane/configure-goal/dry-run. These endpoints disclose control-plane state, goal and run data, local absolute file paths, private Markdown content below registered goal roots, and dry-run configuration projections. The affected routes do not permit cross-origin writes or arbitrary file reads outside registered goal roots.
Root Cause
The root cause is improper origin validation. The server treats loopback binding as a sufficient security boundary and advertises a wildcard CORS policy. Browsers honor that wildcard regardless of the requesting origin, so an attacker-controlled page running in the operator's browser can read responses from the local service.
Attack Vector
Exploitation requires user interaction. The operator must visit attacker-controlled web content while loopx serve-status is running on the same host. The malicious page issues cross-origin GET and POST requests to the loopback port, and the browser returns the response body to attacker JavaScript because of the wildcard ACAO header.
# Security patch in loopx/status_server.py
# fix(status-server): stop sending ACAO:* on unauthenticated read responses (#3137)
def cors_response_headers(origin: str | None) -> dict[str, str]:
"""Return CORS headers for an unauthenticated response.
Only loopback browser origins may read responses cross-origin. A
non-loopback ``Origin`` gets no ``Access-Control-Allow-Origin`` header so
browsers block reads; non-browser clients (no ``Origin`` header) need no
CORS headers at all.
"""
if not origin or not is_loopback_origin(origin):
return {}
return {
"Access-Control-Allow-Origin": origin,
"Access-Control-Allow-Methods": "GET, POST, OPTIONS",
"Access-Control-Allow-Headers": "Content-Type",
"Vary": "Origin",
}
Source: GitHub Commit 838720f
Detection Methods for CVE-2026-108372
Indicators of Compromise
- Unexpected inbound requests to loopback status ports carrying an Origin header from a non-loopback domain.
- Browser-driven GET /status.json, GET /review-material, or POST /control-plane/configure-goal/dry-run requests originating from user-browsing sessions.
- Outbound DNS or HTTPS beacons from workstations shortly after operator browsing activity, which can indicate exfiltration of harvested status data.
Detection Strategies
- Audit the installed LoopX version with loopx --version or pip show loopx across developer and operator endpoints and flag any release below 0.4.5.
- Inspect HTTP response headers on local LoopX ports for Access-Control-Allow-Origin: * using a local request to /status.json.
- Review proxy or endpoint telemetry for browser traffic to 127.0.0.1 or localhost ports associated with LoopX processes.
Monitoring Recommendations
- Enable process and network telemetry for the loopx binary to correlate browser activity with cross-origin reads of the loopback service.
- Monitor goal root directories for anomalous read patterns against private Markdown material.
- Alert on new or unexpected listeners bound to loopback addresses in development environments.
How to Mitigate CVE-2026-108372
Immediate Actions Required
- Upgrade LoopX to version 0.4.5 or later on all systems that run loopx serve-status.
- Stop any running loopx serve-status instances on vulnerable hosts until the upgrade is applied.
- Instruct operators to avoid untrusted web content while the status server is running.
Patch Information
The fix is included in LoopX 0.4.5. The patch introduces a cors_response_headers() helper that returns CORS headers only when the request Origin is a loopback origin, and omits Access-Control-Allow-Origin for all other requests. Review the GitHub Security Advisory GHSA-vx2m-gpq4-8j5q, Pull Request #3137, and Release v0.4.5 for full details.
Workarounds
- Do not run loopx serve-status on hosts where operators browse untrusted content.
- Bind the status server behind a local firewall rule that blocks browser-origin traffic to the loopback port where feasible.
- Use a browser profile dedicated to LoopX workflows that does not visit third-party sites while the status server is active.
# Upgrade LoopX to the patched release
pip install --upgrade "loopx>=0.4.5"
# Verify the installed version
loopx --version
Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
Sources: National Vulnerability Database, CISA KEV, Exploit-DB

