Open WebUI: Terminal proxy forwards a spoofable, integrity-unbound user identity to the upstream (X-User-Id header and ws_terminal session_id query injection)
High8.0CVE-2026-59224 · Published Jul 24, 2026 · updated Aug 4, 2026
Affected versions
| Package | Affected | Fixed in |
|---|---|---|
| open-webui PyPI | < 0.10.0 | 0.10.0 |
Details and references
## Summary The terminal proxy in `backend/open_webui/routers/terminals.py` forwards the Open WebUI user's identity to the upstream terminal server / backend coordinator as an authorization claim, with no cryptographic binding to the session that produced it. The forwarded identity is attacker-influenceable on both proxy paths: 1. **HTTP path (`proxy_terminal`)** sets `headers['X-User-Id'] = user.id`. Upstreams that trust `X-User-Id` as identity receive it unsigned, so an attacker who can reach the upstream by other means (directly, a compromised peer, SSRF) can spoof it. 2. **WebSocket path (`ws_terminal`)** is exploitable through Open WebUI itself, with no "other means" required. It interpolates the path parameter `session_id` directly into the upstream URL and then appends `?user_id=<caller>`: ```python upstream_url = f'{ws_base}/p/{policy_id}/api/terminals/{session_id}' upstream_url += f'?{urllib.parse.urlencode({"user_id": user.id})}' ``` `session_id` is neither validated nor URL-encoded (the HTTP sibling runs `_sanitize_proxy_path`; this path runs nothing). An encoded `?`/`&` smuggled through `session_id` survives Open WebUI's single decode and is re-decoded by the upstream, injecting an attacker-chosen `user_id` ahead of the appended one. Query parsing binds the first occurrence, so the backend coordinator resolves the spoofed user's terminal scope. ## Technical Details The forwarded terminal identity is a bearer-style authorization claim with no integrity binding, and on the WebSocket path it is additionally injectable because `session_id` is concatenated into the URL without encoding or delimiter validation. ## Impact A normal authenticated user can make the terminal proxy present another user's identity to the upstream backend coordinator. On backend coordinator-backed (`policy_id`) servers that scope terminal containers by `user_id`, this reaches another user's terminal scope; combined with a known active session ID (for example a chat-scoped session ID surfaced through a shared chat), it allows attaching to that user's live PTY. The HTTP-path variant additionally allows identity spoofing at the upstream tier for any deployment whose upstream trusts `X-User-Id`. ## Appendix: Affected code - `backend/open_webui/routers/terminals.py` , `proxy_terminal` sets `headers['X-User-Id'] = user.id` with no signature. - `backend/open_webui/routers/terminals.py` , `ws_terminal` builds the upstream URL from an unvalidated, unencoded `session_id` and appends `user_id` as a query parameter, allowing query injection. ## Appendix: Consolidation Per the Report Handling policy, this consolidates independent reports of the same root cause (the forwarded terminal identity is spoofable / not integrity-bound) into the earliest filing: - **@smoke-wolf** (earliest filing) , the `X-User-Id` HTTP-path identity is forwarded without integrity binding, spoofable where the upstream trusts the header. - **@rexpository** , the `ws_terminal` `session_id` query-injection vector, proving the forwarded `user_id` is spoofable through the Open WebUI proxy itself, with no "reach the upstream by other means" precondition. ## Appendix: Recommended fix - Validate and URL-encode `session_id` before building the upstream URL (`urllib.parse.quote(session_id, safe="")`; reject `?`, `#`, `&`, `/`, `%`, backslash, control characters). Build the query string with a URL builder so attacker-controlled path content cannot precede it. - Bind the forwarded identity instead of passing a raw `user_id` / `X-User-Id`: emit a short-lived signed claim (for example HS256 over `{uid, iat, aud:server_id}` with a key shared only with the specific upstream) and verify it upstream.
- CVSS 3.1
- CVSS:3.1/AV:N/AC:H/PR:L/UI:R/S:C/C:H/I:H/A:H
- Severity from
- GitHub (reviewed advisory)
- Weakness
- CWE-287, CWE-290
- Also known as
- CVE-2026-59224, PYSEC-2026-3600
- github.com/open-webui/open-webui/security/advisories/GHSA-j657-m4c4-24jq
- nvd.nist.gov/vuln/detail/CVE-2026-59224
- github.com/open-webui/open-webui/pull/26042
- github.com/open-webui/open-webui/commit/5f3a628a8d291bb5d33e1a0b0c89fb62a2927934
- github.com/open-webui/open-webui
- github.com/open-webui/open-webui/releases/tag/v0.10.0
More Open WebUI advisories
All Open WebUI| Date | Advisory | Severity | Fixed in |
|---|---|---|---|
| Jul 23 | Open WebUI's API key endpoint restrictions bypassed via `x-api-key` header , full message processing on restricted endpoints CVE-2026-45339Medium6.5fixed in 0.9.0 | Medium6.5 | 0.9.0 |
| Jul 24 | Open WebUI: Stored web worker XSS via Pyodide CVE-2026-59214High7.3fixed in 0.10.0 | High7.3 | 0.10.0 |
| Jul 24 | Open WebUI: Account enumeration via observable login timing discrepancy CVE-2026-59218Medium5.3fixed in 0.10.0 | Medium5.3 | 0.10.0 |
| Jul 24 | Open WebUI: Scheduled automations continue after pending-user deactivation and stored model ACL revocation CVE-2026-59226Low3.1fixed in 0.10.0 | Low3.1 | 0.10.0 |
| Jul 24 | Open WebUI: ReDoS in skill-mention regexes causes whole-instance DoS on default config CVE-2026-59220Medium6.5fixed in 0.10.0 | Medium6.5 | 0.10.0 |
| Jul 24 | Open WebUI: POST /api/v1/images/edit bypasses the global image-edit switch and the per-user image-generation permission CVE-2026-59227Medium4.3fixed in 0.10.0 | Medium4.3 | 0.10.0 |