Skip to main content

TencentCloud Octop CVE-2026-108581

| EUVDEUVD-2026-96364 HIGH
Missing Authorization (CWE-862)
2026-10-10 disclosure@vulncheck.com GHSA-c344-c8vm-8rxv
7.1
CVSS 4.0 · Vendor: vulncheck
View

Attack ChainAIDerived

Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain

  1. Access
    Obtain low-privileged authenticated JWT
  2. Exploit
    Query GET /api/providers and /api/voice/providers
  3. Execution
    Receive plaintext LLM/voice provider API keys
  4. Impact
    Abuse upstream provider accounts

Vulnerability Assessment

Weakness
CWE
OWASP Top 10:2025
A01:2025 - Broken Access Control
ASVS 5.0
V8 - Authorization Authorization checks on every request
MITRE ATT&CK

1 technique across 1 tactic: Privilege Escalation

The technique list with ATT&CK links and the kill-chain view is part of the Basic plan and up.

Exploitation Conditions

Requires a valid authenticated low-privileged Octop account/JWT (PR:L) - not exploitable by fully unauthenticated attackers - and requires that one or more LLM or voice provider API keys have actually been configured and stored in the instance. The practical exposure applies to multi-user or multi-tenant deployments where low-privileged users exist who should not have access to provider credentials; in a single-operator install the data is only readable by its legitimate owner. The limiting factor is that the two endpoints still enforce JWT validity (authentication succeeds) but perform no per-role or per-owner authorization check, so any authenticated session is sufficient. No special non-default configuration, platform, or user interaction is needed beyond holding a token.

Impact

This is a genuine, moderately-high-priority confidentiality flaw rather than a CVSS-inflated non-issue, but its real-world urgency is tightly bounded by the authentication requirement. The provided CVSS 4.0 vector (AV:N/AC:L/AT:N/PR:L/UI:N/VC:H/VI:N/VA:N, 7.1 High) is internally consistent with the description: network-reachable endpoints, no user interaction, low attacker effort, and high confidentiality impact (plaintext API key disclosure) with no integrity or availability impact. The key limiting signal is PR:L - exploitation requires a valid authenticated low-privileged account, so this is not a pre-auth mass-exploitable bug. No EPSS score, CISA KEV listing, or formal exploit code is present in the input, so there is NO evidence of active or widespread exploitation; the vulnerability is best treated as an insider/post-compromise privilege-and-data-exposure issue. A detailed public technical writeup (hackmd) and a VulnCheck advisory exist, which raises the practical likelihood that a working reproduction is or will become available even though packaged exploit code is not confirmed. SSVC-style read: exploitation = none observed, automatable = yes (simple authenticated GET requests, no special preconditions beyond a token), technical impact = partial-to-total for the exposed credentials. Net: prioritize for multi-user/multi-tenant Octop deployments where low-privileged accounts exist and provider keys are stored; de-prioritize for single-operator installs.

Exploit ScenarioDerived

An attacker who holds or phishes any low-privileged Octop account (or an authorized but untrusted tenant user) sends authenticated GET requests to /api/providers and /api/voice/providers; because these endpoints validate only the JWT and not the caller's role or ownership, they return the stored LLM and voice provider API keys in plaintext. The attacker then uses those upstream credentials to run up charges, exfiltrate data, or pivot into the connected provider accounts. No packaged public exploit is confirmed, but a public technical writeup describing the cross-role read exists.

This Week Schedule remediation this week. Why: HIGH severity (CVSS 7.0-8.9).
No patch listed yet 0.2% exploitation probability (EPSS)

Get an e-mail when this CVE is exploited, patched or re-scored. Free account.

Alert me

AnalysisAIv1 – Oct 10, 2026

Missing authorization in TencentCloud Octop through 1.0.2b6 lets any authenticated low-privileged user retrieve plaintext LLM and voice provider API keys by calling GET /api/providers and GET /api/voice/providers. Those routes validate the caller's JWT but perform no role, tenant, or ownership check, so in multi-user or multi-tenant deployments any account holding a token is sufficient to harvest credentials and abuse the upstream provider accounts; a single-operator install is largely limited to the legitimate owner reading their own keys. Independent assessment rates this as a genuine moderately-high confidentiality flaw bounded by the PR:L prerequisite (CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:N), and the current exploitation status is: no public exploit identified at time of analysis, despite a public technical writeup and upstream pull requests referencing the issue.

RemediationAI

Patch status: upstream fix available (PR/commit); released patched version not independently confirmed. Track https://github.com/TencentCloud/Octop/pull/1507 and https://github.com/TencentCloud/Octop/pull/1265 and upgrade to the first tagged Octop release that contains them - note the supplied PR 1265 diff is largely unrelated frontend/accessibility work, so confirm the authorization change is actually present in the build you deploy rather than assuming the PR number alone resolves it. Until a patched release is available and deployed, treat every LLM and voice provider API key stored in the instance as compromised: rotate them at the upstream providers, issue replacements with least-privilege scopes and spending limits, and reconfigure Octop with the new keys - the trade-off is brief service interruption for dashboards or agents pinned to the old keys. Reduce the number of authenticated accounts that can reach the API (or disable multi-user access if the deployment allows), since any valid JWT is sufficient; this costs you delegated/team workflows. Restrict network reachability of /api/providers and /api/voice/providers to trusted management networks or an authenticating reverse proxy with an IP allowlist, accepting that remote dashboard users must then tunnel in. If you own the deployment, the direct fix is to add an owner/role authorization dependency to those handlers and mask key material in serialized responses - effective but requires code maintenance and testing on upgrade. Enable or retain access logging on those two routes to detect credential harvesting, and monitor upstream provider usage dashboards for anomalies that indicate keys are already being abused.

Affected VersionsENISA

Octop 0 ≤1.0.2b6

Affected Packages

euvd octop no fix

Source: OSV (Open Source Vulnerabilities)

Affects your stack

Affects 1 package in euvd

Checking this CVE against your own stacks and servers is part of the Basic plan and up.

Technical ContextAI · show details

Octop is a Python/FastAPI-based LLM and voice agent orchestration platform whose dashboard is backed by JSON routers, including the provider router at src/octop/api/routers/providers.py (referenced lines 120-208) and an analogous /api/voice/providers route. The root-cause class is CWE-862 (Missing Authorization): the routes are guarded by a JWT validation dependency that confirms the caller is a valid, authenticated user, but no downstream dependency or in-handler check scopes the returned provider records to the caller's role, tenant, or ownership. Because provider configuration objects hold the upstream LLM and voice provider API keys and are serialized back in plaintext without masking, the endpoints effectively function as a cross-role credential read primitive - the reporter's 'Authentication Bypass' tag reflects the practical outcome (authorization bypass) even though authentication itself still succeeds. This is a conflation of authentication with authorization rather than a token-validation weakness, and it needs no non-default configuration, platform, or user interaction beyond possessing a valid token and at least one stored provider key. No CPE strings were supplied with the advisory data, so identification relies on the vendor/product name and source references.

Know when the next one hits your stack

  • Watch products and get an email when something new is published
  • Check this CVE against your own stack and servers
  • Exploit and PoC references, threat intelligence from 22 sources

No credit card · 7-day full trial

Severity by source

Vendor (vulncheck) PRIMARY 7.1 HIGH
CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:H/VI:N/VA:N/SC:N/SI:N/SA:N
vuln.today AI 6.5 MEDIUM

Network-reachable endpoints need only a valid low-privileged JWT (PR:L, AC:L, no UI); impact is plaintext credential disclosure so C:H, I:N, A:N, scope unchanged.

3.1 AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:N
4.0 AV:N/AC:L/AT:N/PR:L/UI:N/VC:H/VI:N/VA:N/SC:N/SI:N/SA:N

Primary rating from Vendor (vulncheck).

CVSS VectorVendor: vulncheck

Attack Vector
Network
Attack Complexity
Low
Attack Requirements
None
Privileges Required
Low
User Interaction
None
Vulnerable system: Confidentiality
High
Vulnerable system: Integrity
None
Vulnerable system: Availability
None
Subsequent system: Confidentiality
None
Subsequent system: Integrity
None
Subsequent system: Availability
None

DescriptionCVE.org

TencentCloud Octop through 1.0.2b6 contains a missing authorization vulnerability that allows authenticated low-privileged users to read stored provider API keys via GET /api/providers and GET /api/voice/providers. Attackers can query these endpoints, which only validate the JWT, to obtain plaintext LLM and voice provider API keys and abuse the upstream provider accounts.

Executive ViewAI

Why It Matters

TencentCloud Octop deployments up to version 1.0.2b6 expose stored LLM and voice provider API keys to any user holding a valid login token, because two API endpoints return credentials without checking role, tenant, or ownership. In multi-user or multi-tenant environments this lets a low-privileged insider or any compromised standard account harvest billable third-party credentials, creating direct financial exposure and downstream service abuse. Leadership should treat this as a credential-governance and cloud-spend risk rather than a purely technical bug, since the leaked keys are the keys to externally billed AI and voice services.

Business Impact: High
Compensating Controls

No vendor-released patch identified at time of analysis; interim mitigations: rotate and scope all stored LLM/voice provider keys now, restrict or block network access to the affected API routes, reduce Octop to single-operator or strictly administrator-only accounts, add upstream spend limits and anomaly alerting, and place an authorization-enforcing reverse proxy in front of Octop.

Threat Intelligence Overview

EPSS Score
0.2%
Top 86% exploitable
Prevalence
Low
niche self-hosted LLM/voice agent platform, limited production footprint
Social Media
1
Sightings
Data Quality
6 sources
Confidence: High

How this decision was reached

The engine reads each signal below and applies a fixed list of rules from top to bottom. The first rule that matches sets the recommended action – like a statement, every line is shown.

SignalValue
Severity HIGH
CISA KEV No
Public POC No
EPSS 0.2%
Patch available No
Active threat No
Deciding rule
HIGH severity (CVSS 7.0-8.9)
What every action level means
Emergency Immediate action required. Criteria: CRITICAL + in CISA KEV (actively exploited) without patch, OR CRITICAL + EPSS > 30% without patch, OR CRITICAL + active threat signals without patch.
Act Now Prioritize within 24-48h. Criteria: in KEV with patch available, OR CRITICAL + public POC, OR HIGH + EPSS > 10%, OR CRITICAL + active threat, OR any CRITICAL severity.
This Week Schedule remediation this week. Criteria: HIGH severity (CVSS 7.0-8.9). Patch within normal change window.
This Month Plan within 30 days. Criteria: MEDIUM severity (CVSS 4.0-6.9), lower exploitation probability. Include in regular patch cycle.
Monitor Track and reassess. Criteria: LOW severity (CVSS 0.1-3.9), minimal exploitation risk. No immediate action needed.

SSVC AssessmentENISA

SSVC Decision· derived
Scheduled
Exploitation· derived
None
System Exposure· derived
Open
Automatable· derived
No
Human Impact
High

Priority Score

36
Low Medium High Critical
KEV: 0
EPSS: 0
CVSS: +36
PoC: 0
Weaponized: 0
Threat intel: 0
Gov advisory: 0

Attack Technique

Authentication bypass attacks exploit flaws in the verification mechanisms that control access to systems and applications.

This site uses cookies essential for authentication and security. No tracking or analytics cookies are used. Privacy Policy