Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain
-
AccessReach exposed Verba generate_stream endpoint
-
ExploitCall vulnerable get_environment function
-
ImpactHarvest leaked environment/configuration data
Vulnerability Assessment
| Weakness |
CWE
OWASP Top 10:2025
A01:2025 - Broken Access Control
ASVS 5.0
V14 - Data Protection Data exposure prevention
MITRE ATT&CK
2 techniques across 2 tactics: Collection, Discovery The technique list with ATT&CK links and the kill-chain view is part of the Basic plan and up. |
| Exploitation Conditions | Remote, unauthenticated exploitation (PR:N, UI:N) against the generate_stream endpoint of a reachable Weaviate Verba instance running version 2.1.3 or earlier; triggering the get_environment function returns environment/configuration information to the caller. The practical limiting factor is network reachability - the Verba service must be exposed to the attacker (e.g., deployed without a front-end authentication/network control) - and impact is bounded to confidentiality only (VC:L / C:L), meaning partial information disclosure rather than code execution or data modification. No special non-default feature toggle is indicated by the description beyond having the vulnerable endpoint accessible. |
| Impact | This is a moderate-severity information-disclosure flaw (CWE-200) in Weaviate Verba, an open-source RAG application. The provided CVSS 4.0 vector (AV:N/AC:L/AT:N/PR:N/UI:N/VC:L/VI:N/VA:N) describes a remotely reachable, low-complexity, unauthenticated issue with only LOW confidentiality impact and no integrity or availability impact, yielding a base score of 6.9. The exploit-maturity metric E:P (Proof-of-Concept) is consistent with the reported publicly available POC (a gist). No EPSS score is provided and the CVE is NOT listed in CISA KEV, so there is no evidence of active, widespread exploitation - exploitation status is limited to publicly available exploit code. Signals are largely consistent: easy remote access and public POC raise the practical likelihood of opportunistic probing, but the impact ceiling is modest (partial information disclosure, VC:L, no write/destructive effect). The notable gap is vendor response: VulDB states the vendor was contacted but did not respond, so there is no vendor-confirmed patch or advisory, which increases uncertainty about fix availability and precise affected-version boundaries. Net: treat as a real but non-emergency priority - worth patching/mitigating promptly on any internet-exposed Verba instance, but not a drop-everything issue given low impact, niche deployment footprint, and absence of KEV/EPSS escalation signals. |
| Exploit ScenarioDerived | An attacker who can reach a network-exposed Verba instance sends a request to the generate_stream endpoint, which invokes the vulnerable get_environment function in goldenverba/components/util.py and returns environment/configuration information back to the unauthenticated caller. A public proof-of-concept exploit exists (published as a gist), lowering the skill barrier for opportunistic attackers. Because the CVSS vector is AV:N/AC:L/PR:N/UI:N, no credentials, user interaction, or special timing are required. |
Get an e-mail when this CVE is exploited, patched or re-scored. Free account.
Alert meAnalysisAIv1 – Oct 04, 2026
Information disclosure in Weaviate Verba 2.1.3 and earlier allows remote, unauthenticated attackers to retrieve environment and configuration data by invoking the get_environment function in goldenverba/components/util.py through the generate_stream endpoint. Publicly available exploit code exists (referenced in the VulDB disclosure and a public gist), raising the likelihood of opportunistic scanning, although there is no confirmation of active exploitation (not listed in CISA KEV). Impact is bound to confidentiality (partial disclosure), yielding a 6.9 CVSS v4.0 base score; the main practical prerequisite is that the Verba service be network-reachable.
RemediationAI
No vendor-released patch has been identified at time of analysis; the disclosure states the vendor was contacted early but did not respond, so no fixed version number can be cited. Until an official fix ships, apply compensating controls that reflect the actual exposure path: remove the Verba service from untrusted networks and place it behind an authenticating reverse proxy or identity-aware gateway so that the generate_stream endpoint cannot be reached anonymously - the trade-off is that any existing API clients or embedded chat widgets must be updated to send credentials. If the streaming feature is not required, block the /generate_stream route at the reverse proxy or WAF entirely; this is effective but breaks the application's chat streaming for legitimate users. Because the leaked data is process environment content, treat every credential stored in the Verba runtime environment (model-provider API keys, database passwords, vector-store tokens) as potentially compromised and rotate them, and reduce the secrets present in that environment going forward - note that rotation is disruptive but necessary if the endpoint was reachable from an untrusted network. Finally, monitor and alert on requests to generate_stream from unexpected source addresses, and review the VulDB advisories (https://vuldb.com/vuln/413370, https://vuldb.com/cve/CVE-2026-105145) for an eventual patched release; upgrade to that release once it exists.
Affected VersionsENISA
Affected Packages
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
Weaviate Verba is an open-source retrieval-augmented generation (RAG) chat application built on top of the Weaviate vector database; its GoldenVerba front end handles document ingestion, chunking, and streaming chat responses. The flaw sits in the get_environment helper of goldenverba/components/util.py, which reads and returns process environment information. Because this helper is reachable through the generate_stream endpoint without an authentication or authorization check, it exposes CWE-200 sensitive information - typically environment variables that hold OpenAI or other model-provider API keys, database connection strings, and deployment configuration. The vulnerability class (exposure of sensitive information to an unauthorized actor) is a design and access-control weakness rather than memory corruption: the attacker simply causes the endpoint to serialize internal configuration back to the caller. The CPE cpe:2.3:a:weaviate:verba:*:*:*:*:*:*:*:* confirms the affected package is the Weaviate Verba application itself, and EUVD enumerates builds 2.1.0 through 2.1.3 as vulnerable.
Know when Weaviate gets its next vulnerability
- 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
CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:L/VI:N/VA:N/SC:N/SI:N/SA:N/E:P
Remotely reachable endpoint with no auth or interaction (AV:N/AC:L/PR:N/UI:N); only partial information disclosure, so C:L with I:N/A:N and unchanged scope.
Primary rating from Vendor (VulDB).
CVSS VectorVendor: VulDB
Lifecycle Timeline
2DescriptionCVE.org
A vulnerability has been found in Weaviate Verba up to 2.1.3. Affected by this vulnerability is the function get_environment of the file goldenverba/components/util.py of the component generate_stream Endpoint. The manipulation leads to information disclosure. It is possible to initiate the attack remotely. The exploit has been disclosed to the public and may be used. The vendor was contacted early about this disclosure but did not respond in any way.
Threat Intelligence Overview
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.
What every action level means
SSVC AssessmentENISA
Priority Score
Attack Technique
Information disclosure occurs when an application unintentionally exposes sensitive data that aids attackers in reconnaissance or directly compromises security.
An issue in weaviate v.1.20.0 allows a remote attacker to cause a denial of service via the handleUnbatchedGraphQLReques
Server-side request forgery in Weaviate's Verba (GoldenVerba) RAG application version 2.1.3 lets unauthenticated remote
Privilege escalation in Weaviate vector database versions before 1.38.0 allows a user holding only the delegated assign_
Path traversal in Weaviate OSS before 1.33.4 allows a high-privileged user who can insert data into the database to craf
Authorization bypass in Weaviate's Static API Key Handler (versions 1.37.0-1.37.7) stems from the validateConfig functio
An arbitrary file read flaw in Weaviate OSS before 1.33.4 lets an attacker with high privileges (PR:H, authenticated wit
EUVD-2026-92090
GHSA-8456-r8c8-j656
NVD
CVE.org
EPSS
Vulnrichment
VulDB
Exploit-DB