π¨ CVE-2026-18813
A vulnerability has been found in H3C NX15 V100R017. This affects the function delete of the file /api/esps. The manipulation of the argument esps.apcm.version leads to command injection. The attack can be initiated remotely. The exploit has been disclosed to the public and may be used. The vendor was contacted early about this disclosure.
π@cveNotify
A vulnerability has been found in H3C NX15 V100R017. This affects the function delete of the file /api/esps. The manipulation of the argument esps.apcm.version leads to command injection. The attack can be initiated remotely. The exploit has been disclosed to the public and may be used. The vendor was contacted early about this disclosure.
π@cveNotify
GitHub
IOT_Vul_Public/H3C/NX15R017/esps_apcm_version_delete_root_rce at main Β· coconut652-7/IOT_Vul_Public
some PoCs for CVE submissions. Contribute to coconut652-7/IOT_Vul_Public development by creating an account on GitHub.
π¨ CVE-2026-54020
Open WebUI is an extensible, feature-rich, and user-friendly self-hosted AI platform. Prior to 0.11.0, Open WebUI resolved a hostname during URL validation and rejected private, loopback, and link-local addresses, but the HTTP clients resolved the hostname again at connection time. An authenticated attacker who controlled authoritative DNS for a submitted hostname could answer with a public address during validation and an internal one during connection, reaching cloud metadata, loopback admin APIs, or internal services through URL ingest, chat image_url fetches, image editing, or OAuth profile-picture fetches, with most paths returning the response to the attacker and the OAuth path forwarding the OAuth access token. This issue is fixed in 0.11.0.
π@cveNotify
Open WebUI is an extensible, feature-rich, and user-friendly self-hosted AI platform. Prior to 0.11.0, Open WebUI resolved a hostname during URL validation and rejected private, loopback, and link-local addresses, but the HTTP clients resolved the hostname again at connection time. An authenticated attacker who controlled authoritative DNS for a submitted hostname could answer with a public address during validation and an internal one during connection, reaching cloud metadata, loopback admin APIs, or internal services through URL ingest, chat image_url fetches, image editing, or OAuth profile-picture fetches, with most paths returning the response to the attacker and the OAuth path forwarding the OAuth access token. This issue is fixed in 0.11.0.
π@cveNotify
GitHub
Release v0.11.0 Β· open-webui/open-webui
Added
π¨ Redesigned interface. Open WebUI has been visually rebuilt from the ground up. All aspects of the User Interface, from the chat view to the admin panel. Now with a narrower conversation co...
π¨ Redesigned interface. Open WebUI has been visually rebuilt from the ground up. All aspects of the User Interface, from the chat view to the admin panel. Now with a narrower conversation co...
π¨ CVE-2026-70488
Open WebUI is an extensible, feature-rich, and user-friendly self-hosted AI platform. From 0.9.6 until 0.11.0, the sync cleanup endpoint authorized write access to the knowledge base in the URL but then acted on directory and file ids supplied in the request body without checking that those objects belonged to that knowledge base. A user with write access to one knowledge base could delete directories and remove file embeddings from another knowledge base, causing documents to drop out of retrieval results and breaking chat-with-file for targeted documents without disclosing contents. This issue is fixed in 0.11.0.
π@cveNotify
Open WebUI is an extensible, feature-rich, and user-friendly self-hosted AI platform. From 0.9.6 until 0.11.0, the sync cleanup endpoint authorized write access to the knowledge base in the URL but then acted on directory and file ids supplied in the request body without checking that those objects belonged to that knowledge base. A user with write access to one knowledge base could delete directories and remove file embeddings from another knowledge base, causing documents to drop out of retrieval results and breaking chat-with-file for targeted documents without disclosing contents. This issue is fixed in 0.11.0.
π@cveNotify
GitHub
fix: scope knowledge sync cleanup deletions to the target knowledge b⦠· open-webui/open-webui@707efea
β¦ase (#26722)
POST /knowledge/{id}/sync/cleanup verified write access to the knowledge base in the URL but then acted on the caller-supplied file_ids and dir_ids without checking they belong to th...
POST /knowledge/{id}/sync/cleanup verified write access to the knowledge base in the URL but then acted on the caller-supplied file_ids and dir_ids without checking they belong to th...
π¨ CVE-2026-70489
Open WebUI is an extensible, feature-rich, and user-friendly self-hosted AI platform. From 0.9.0 until 0.11.0, automation recurrence parsing in backend/open_webui/utils/automations.py anchored minutely and hourly rules at a fixed date of 2000-01-01 and then walked forward one interval at a time to find the next run. A single FREQ=MINUTELY rule enumerates roughly a quarter-century of occurrences synchronously on the event loop that also serves scheduler, HTTP, and WebSocket traffic, and the scheduler recomputes the next run for every claimed row on each poll. This causes availability impact for every other user of the instance. This issue is fixed in 0.11.0.
π@cveNotify
Open WebUI is an extensible, feature-rich, and user-friendly self-hosted AI platform. From 0.9.0 until 0.11.0, automation recurrence parsing in backend/open_webui/utils/automations.py anchored minutely and hourly rules at a fixed date of 2000-01-01 and then walked forward one interval at a time to find the next run. A single FREQ=MINUTELY rule enumerates roughly a quarter-century of occurrences synchronously on the event loop that also serves scheduler, HTTP, and WebSocket traffic, and the scheduler recomputes the next run for every claimed row on each poll. This causes availability impact for every other user of the instance. This issue is fixed in 0.11.0.
π@cveNotify
GitHub
refac Β· open-webui/open-webui@c4ae8c8
User-friendly AI Interface (Supports Ollama, OpenAI API, ...) - refac Β· open-webui/open-webui@c4ae8c8
π¨ CVE-2026-70493
Open WebUI is an extensible, feature-rich, and user-friendly self-hosted AI platform. From 0.9.6 until 0.11.0, the built-in knowledge search path in backend/open_webui/tools/knowledge_fs.py and backend/open_webui/tools/builtin.py let a chat participant choose a pattern used to grep knowledge files. Patterns containing regex metacharacters were compiled with Python's backtracking re engine and run against every line of every reachable file with no time limit, so a crafted pattern such as (x|x)*y and one matching uploaded file line can pin one CPU core and block the event loop. This causes availability impact for every other user of the affected worker. This issue is fixed in 0.11.0.
π@cveNotify
Open WebUI is an extensible, feature-rich, and user-friendly self-hosted AI platform. From 0.9.6 until 0.11.0, the built-in knowledge search path in backend/open_webui/tools/knowledge_fs.py and backend/open_webui/tools/builtin.py let a chat participant choose a pattern used to grep knowledge files. Patterns containing regex metacharacters were compiled with Python's backtracking re engine and run against every line of every reachable file with no time limit, so a crafted pattern such as (x|x)*y and one matching uploaded file line can pin one CPU core and block the event loop. This causes availability impact for every other user of the affected worker. This issue is fixed in 0.11.0.
π@cveNotify
GitHub
fix: bound knowledge-search matching so one pattern cannot stall the β¦ Β· open-webui/open-webui@3ab2026
β¦worker (#27471)
build_matcher compiled a caller-supplied pattern with Python's backtracking re and ran it over every line of every reachable file, with no timeout, no thread offload and n...
build_matcher compiled a caller-supplied pattern with Python's backtracking re and ran it over every line of every reachable file, with no timeout, no thread offload and n...
π¨ CVE-2026-70494
Open WebUI is an extensible, feature-rich, and user-friendly self-hosted AI platform. From 0.10.0 until 0.11.0, the DELETE /api/v1/folders/{id} handler in backend/open_webui/routers/folders.py allowed a user granted write access to a shared chat folder to permanently delete chats and messages belonging to the folder owner. The cascade following the authorization check is bound to the folder owner's id, but the subfolder check accepted any inherited write grant instead of requiring ownership or administrator status. A collaborator can destroy the owner's subtree or force-move chats out of it when delete_contents=false. This issue is fixed in 0.11.0.
π@cveNotify
Open WebUI is an extensible, feature-rich, and user-friendly self-hosted AI platform. From 0.10.0 until 0.11.0, the DELETE /api/v1/folders/{id} handler in backend/open_webui/routers/folders.py allowed a user granted write access to a shared chat folder to permanently delete chats and messages belonging to the folder owner. The cascade following the authorization check is bound to the folder owner's id, but the subfolder check accepted any inherited write grant instead of requiring ownership or administrator status. A collaborator can destroy the owner's subtree or force-move chats out of it when delete_contents=false. This issue is fixed in 0.11.0.
π@cveNotify
GitHub
fix: restrict folder deletion to the owner or an admin (#27003) Β· open-webui/open-webui@915ef7d
* fix: restrict folder deletion to the owner or an admin
Deleting a folder cascades into the folder owner's chats, messages and the
entire subfolder subtree; the cascade is bound to the fo...
Deleting a folder cascades into the folder owner's chats, messages and the
entire subfolder subtree; the cascade is bound to the fo...
π¨ CVE-2026-70554
MaxSite CMS contains a PHP object injection vulnerability that allows unauthenticated attackers to execute arbitrary code by passing attacker-controlled serialized data in the maxsite_comuser cookie directly to unserialize() without validation or class allowlisting. Attackers can craft a malicious serialized PHP object payload delivered in a single HTTP request to trigger magic methods during object graph reconstruction, enabling property-oriented programming attacks or remote code execution via available gadget chains such as those targeting SoapClient or Imagick extensions.
π@cveNotify
MaxSite CMS contains a PHP object injection vulnerability that allows unauthenticated attackers to execute arbitrary code by passing attacker-controlled serialized data in the maxsite_comuser cookie directly to unserialize() without validation or class allowlisting. Attackers can craft a malicious serialized PHP object payload delivered in a single HTTP request to trigger magic methods during object graph reconstruction, enabling property-oriented programming attacks or remote code execution via available gadget chains such as those targeting SoapClient or Imagick extensions.
π@cveNotify
GitHub
GitHub - maxsite/cms: MaxSite CMS. Free CMS for your website. Smart alternative to WordPress
MaxSite CMS. Free CMS for your website. Smart alternative to WordPress - maxsite/cms
π¨ CVE-2026-18814
A vulnerability was found in H3C NX15 V100R017. This impacts the function reload.reload_config of the file /api/esps. The manipulation results in command injection. The attack can be launched remotely. The exploit has been made public and could be used. The vendor was contacted early about this disclosure.
π@cveNotify
A vulnerability was found in H3C NX15 V100R017. This impacts the function reload.reload_config of the file /api/esps. The manipulation results in command injection. The attack can be launched remotely. The exploit has been made public and could be used. The vendor was contacted early about this disclosure.
π@cveNotify
GitHub
IOT_Vul_Public/H3C/NX15R017/service_add_root_rce at main Β· coconut652-7/IOT_Vul_Public
some PoCs for CVE submissions. Contribute to coconut652-7/IOT_Vul_Public development by creating an account on GitHub.
π¨ CVE-2026-18816
A vulnerability was identified in Baserow up to 2.3.2. Affected by this vulnerability is the function verify of the file backend/src/baserow/api/two_factor_auth/views.py of the component 2FA Verify Endpoint. Such manipulation leads to improper authentication. The attack may be launched remotely. This attack is characterized by high complexity. The exploitation appears to be difficult. Upgrading to version 2.3.3 addresses this issue. You should upgrade the affected component. The vendor was contacted early, responded in a very professional manner and quickly released a fixed version of the affected product.
π@cveNotify
A vulnerability was identified in Baserow up to 2.3.2. Affected by this vulnerability is the function verify of the file backend/src/baserow/api/two_factor_auth/views.py of the component 2FA Verify Endpoint. Such manipulation leads to improper authentication. The attack may be launched remotely. This attack is characterized by high complexity. The exploitation appears to be difficult. Upgrading to version 2.3.3 addresses this issue. You should upgrade the affected component. The vendor was contacted early, responded in a very professional manner and quickly released a fixed version of the affected product.
π@cveNotify
GitHub
GitHub - baserow/baserow: Build databases, automations, apps & agents with AI β no code. Open source platform available on cloudβ¦
Build databases, automations, apps & agents with AI β no code. Open source platform available on cloud and self-hosted. GDPR, HIPAA, SOC 2 compliant. Best Airtable alternative. - baserow/baserow
π¨ CVE-2026-45084
OpenSIPS is a Session Initiation Protocol (SIP) server implementation. Versions 3.4.0 through 3.6.5 contain a denial of service vulnerability in the presence module. When the presence module's handle_publish() function processes a SIP PUBLISH request with an Event: presence header and a message body while the configuration option enable_sphere_check=1 is set, it invokes the get_content_type() macro without first calling parse_content_type_hdr(), causing it to dereference uninitialized or NULL Content-Type parsing state and crash. If a Content-Type header is present but unparsed, msg->content_type->parsed is NULL and is dereferenced as a content_t pointer; if the request lacks a Content-Type header entirely, msg->content_type itself is NULL, and both cases lead to a crash. A remote attacker can therefore cause a denial of service against an affected instance with a single PUBLISH request over UDP or TCP, using either a valid Content-Type: application/pidf+xml request or one with the header removed, and the vulnerable code path itself does not enforce authentication (though a deployment's routing configuration may require it before this route is reached). The issue has been fixed in version 3.6.6 and 4.0.0-rc1.
π@cveNotify
OpenSIPS is a Session Initiation Protocol (SIP) server implementation. Versions 3.4.0 through 3.6.5 contain a denial of service vulnerability in the presence module. When the presence module's handle_publish() function processes a SIP PUBLISH request with an Event: presence header and a message body while the configuration option enable_sphere_check=1 is set, it invokes the get_content_type() macro without first calling parse_content_type_hdr(), causing it to dereference uninitialized or NULL Content-Type parsing state and crash. If a Content-Type header is present but unparsed, msg->content_type->parsed is NULL and is dereferenced as a content_t pointer; if the request lacks a Content-Type header entirely, msg->content_type itself is NULL, and both cases lead to a crash. A remote attacker can therefore cause a denial of service against an affected instance with a single PUBLISH request over UDP or TCP, using either a valid Content-Type: application/pidf+xml request or one with the header removed, and the vulnerable code path itself does not enforce authentication (though a deployment's routing configuration may require it before this route is reached). The issue has been fixed in version 3.6.6 and 4.0.0-rc1.
π@cveNotify
GitHub
denial of service in `presence.handle_publish()` from unchecked Content-Type state
## Description
A denial of service vulnerability exists in the `presence` module of OpenSIPS. When the `handle_publish` function processes a SIP `PUBLISH` request with an `Event: presence` heade...
A denial of service vulnerability exists in the `presence` module of OpenSIPS. When the `handle_publish` function processes a SIP `PUBLISH` request with an `Event: presence` heade...
π¨ CVE-2026-45100
OpenSIPS is a Session Initiation Protocol (SIP) server implementation. Versions 3.4.0-beta through 3.6.5 and 4.0.0-beta contain a buffer overflow in the {s.b64encode} string transformation. The size check for {s.b64encode} only verifies that the input fits within the 64 KB transformation buffer, but base64 encoding expands the data by roughly a third, so an input between about 49,153 and 65,535 bytes produces more output than the buffer can hold and overflows it by up to 21,844 bytes. Because these transformation buffers sit next to each other in memory and are reused for chained transformations, the overflow writes attacker-controlled data into the adjacent buffer and corrupts values used by later transformations processing the same SIP message. A remote attacker can trigger this by sending a SIP message with a large header value (roughly 50,000 bytes or more) when the routing script applies {s.b64encode} to attacker-controlled input, making exploitability dependent on the deployment's routing configuration. This issue has been fixed in versions 3.6.6 and 4.0.0-rc1.
π@cveNotify
OpenSIPS is a Session Initiation Protocol (SIP) server implementation. Versions 3.4.0-beta through 3.6.5 and 4.0.0-beta contain a buffer overflow in the {s.b64encode} string transformation. The size check for {s.b64encode} only verifies that the input fits within the 64 KB transformation buffer, but base64 encoding expands the data by roughly a third, so an input between about 49,153 and 65,535 bytes produces more output than the buffer can hold and overflows it by up to 21,844 bytes. Because these transformation buffers sit next to each other in memory and are reused for chained transformations, the overflow writes attacker-controlled data into the adjacent buffer and corrupts values used by later transformations processing the same SIP message. A remote attacker can trigger this by sending a SIP message with a large header value (roughly 50,000 bytes or more) when the routing script applies {s.b64encode} to attacker-controlled input, making exploitability dependent on the deployment's routing configuration. This issue has been fixed in versions 3.6.6 and 4.0.0-rc1.
π@cveNotify
GitHub
core: enforce bounds checks on input-derived lengths (#3888) Β· OpenSIPS/opensips@4d23613
- transformations: account for base64 4/3 expansion in b64encode
output length check
- parser/parse_body: validate remaining buffer length before delimiter
comparison in multipart boundary sear...
output length check
- parser/parse_body: validate remaining buffer length before delimiter
comparison in multipart boundary sear...
π¨ CVE-2026-51144
Cross Site Scripting vulnerability in Soliton Systems MailZen Management Protal v.2.62, v.2.63 allows a remote attacker to execute arbitrary code via the Role Name, First Name, Last Name, and Username fields.
π@cveNotify
Cross Site Scripting vulnerability in Soliton Systems MailZen Management Protal v.2.62, v.2.63 allows a remote attacker to execute arbitrary code via the Role Name, First Name, Last Name, and Username fields.
π@cveNotify
GitHub
Vulnerability-Research/soliton-mailzen.md at main Β· FahadAljuaid/Vulnerability-Research
Contribute to FahadAljuaid/Vulnerability-Research development by creating an account on GitHub.
π¨ CVE-2026-67857
open62541 1.5.5 contains an out-of-bounds read in the client-side function responseReadNamespacesArray() in src/client/ua_client_connect.c.
π@cveNotify
open62541 1.5.5 contains an out-of-bounds read in the client-side function responseReadNamespacesArray() in src/client/ua_client_connect.c.
π@cveNotify
GitHub
`client out-of-bounds` read in `responseReadNamespacesArray()` when processing a zero-length `ReadResponse.results` array encodedβ¦
Summary A malicious OPC UA TCP server can crash the open62541 client during the normal connection handshake after ActivateSession, when the client automatically reads Server_NamespaceArray. The iss...
π¨ CVE-2026-67858
Buffer Overflow vulnerability exists in open62541 1.5.5 when the Local Discovery Server (LDS) is built with multicast discovery enabled through the MDNSD backend. An unauthenticated remote attacker can send a RegisterServer or RegisterServer2 request containing many unique discoveryUrls. This allows remote attackers to cause a denial of service.
π@cveNotify
Buffer Overflow vulnerability exists in open62541 1.5.5 when the Local Discovery Server (LDS) is built with multicast discovery enabled through the MDNSD backend. An unauthenticated remote attacker can send a RegisterServer or RegisterServer2 request containing many unique discoveryUrls. This allows remote attackers to cause a denial of service.
π@cveNotify
GitHub
open62541/doc/building.rst at master Β· open62541/open62541
Open source implementation of OPC UA (OPC Unified Architecture) aka IEC 62541 licensed under Mozilla Public License v2.0 - open62541/open62541
π¨ CVE-2026-70589
Ghost is a Node.js content management system. From 4.22.0 until 6.54.1, a missing validation check allowed users to redeem subscription offers that were no longer active. This issue is fixed in version 6.54.1.
π@cveNotify
Ghost is a Node.js content management system. From 4.22.0 until 6.54.1, a missing validation check allowed users to redeem subscription offers that were no longer active. This issue is fixed in version 6.54.1.
π@cveNotify
GitHub
Fixed offer redemption to reject inactive offers (#29630) Β· TryGhost/Ghost@d91c0fc
no ref
Co-authored-by: UserExistsError <23325451+UserExistsError@users.noreply.github.com>
Co-authored-by: UserExistsError <23325451+UserExistsError@users.noreply.github.com>
π¨ CVE-2026-70590
Ghost is a Node.js content management system. Prior to 6.54.1, any staff-level user was able to leak the hashed passwords of other staff users through the Ghost Admin API. An offline password-guessing attack against the hashes could lead to account takeover if successful, but Device Verification should have prevented an attacker from logging in with a recovered password. Depending on the database used, leaked hashes may not have had the correct casing for all characters, increasing the difficulty of a password-guessing attack. This issue is fixed in version 6.54.1.
π@cveNotify
Ghost is a Node.js content management system. Prior to 6.54.1, any staff-level user was able to leak the hashed passwords of other staff users through the Ghost Admin API. An offline password-guessing attack against the hashes could lead to account takeover if successful, but Device Verification should have prevented an attacker from logging in with a recovered password. Depending on the database used, leaked hashes may not have had the correct casing for all characters, increasing the difficulty of a password-guessing attack. This issue is fixed in version 6.54.1.
π@cveNotify
GitHub
Applied NQL transformer functions to more APIs. (#29628) Β· TryGhost/Ghost@63c31fa
no ref
- this ensures more consistent behavior when using filters.
Co-authored-by: UserExistsError <23325451+UserExistsError@users.noreply.github.com>
- this ensures more consistent behavior when using filters.
Co-authored-by: UserExistsError <23325451+UserExistsError@users.noreply.github.com>
π¨ CVE-2026-70591
Ghost is a Node.js content management system. From 0.10.0 until 6.54.1, a Server-Side Request Forgery in Ghost Admin image fetching allowed any staff-level user to perform a blind HTTP GET request against internal hosts. No output was returned, but this could have been used to probe open ports on internal hosts. This issue is fixed in version 6.54.1.
π@cveNotify
Ghost is a Node.js content management system. From 0.10.0 until 6.54.1, a Server-Side Request Forgery in Ghost Admin image fetching allowed any staff-level user to perform a blind HTTP GET request against internal hosts. No output was returned, but this could have been used to probe open ports on internal hosts. This issue is fixed in version 6.54.1.
π@cveNotify
GitHub
Fixed `got` client inconsistencies across image types Β· TryGhost/Ghost@5eff2de
no ref
- this ensures the same URL host validation is applied when fetching image size.
- this ensures the same URL host validation is applied when fetching image size.
π¨ CVE-2026-70592
Ghost is a Node.js content management system. From 1.20.1 until 6.54.1, an Administrator-level user could remotely overwrite certain files on the filesystem through the database backup filename, leading to integrity and availability issues. The database export endpoint failed to reject path separators in the caller-supplied filename. This issue is fixed in version 6.54.1.
π@cveNotify
Ghost is a Node.js content management system. From 1.20.1 until 6.54.1, an Administrator-level user could remotely overwrite certain files on the filesystem through the database backup filename, leading to integrity and availability issues. The database export endpoint failed to reject path separators in the caller-supplied filename. This issue is fixed in version 6.54.1.
π@cveNotify
GitHub
Improved validation of export filenames Β· TryGhost/Ghost@f466c30
no ref
π¨ CVE-2026-70593
Ghost is a Node.js content management system. From 0.10.0 until 6.54.1, a vulnerability in custom themes allowed a staff user to write files outside of the uploads directory. This could be used to alter the behavior of the installation through custom theme upload path traversal in LocalStorageBase and theme storage name handling. This issue is fixed in version 6.54.1.
π@cveNotify
Ghost is a Node.js content management system. From 0.10.0 until 6.54.1, a vulnerability in custom themes allowed a staff user to write files outside of the uploads directory. This could be used to alter the behavior of the installation through custom theme upload path traversal in LocalStorageBase and theme storage name handling. This issue is fixed in version 6.54.1.
π@cveNotify
GitHub
Improved validation of uploaded theme names Β· TryGhost/Ghost@fbaa923
Independent technology for modern publishing, memberships, subscriptions and newsletters. - Improved validation of uploaded theme names Β· TryGhost/Ghost@fbaa923
π¨ CVE-2026-70594
Ghost is a Node.js content management system. From 2.2.0 until 6.54.1, Ghost Admin did not invalidate existing sessions on login which could have allowed for session fixation attacks. Successful exploitation would have required another vulnerability on the same domain where Ghost Admin was hosted. This issue is fixed in version 6.54.1.
π@cveNotify
Ghost is a Node.js content management system. From 2.2.0 until 6.54.1, Ghost Admin did not invalidate existing sessions on login which could have allowed for session fixation attacks. Successful exploitation would have required another vulnerability on the same domain where Ghost Admin was hosted. This issue is fixed in version 6.54.1.
π@cveNotify
GitHub
Fixed auth session to generate a new admin session on login (#29634) Β· TryGhost/Ghost@6b1c85c
no ref
Co-authored-by: UserExistsError <23325451+UserExistsError@users.noreply.github.com>
Co-authored-by: UserExistsError <23325451+UserExistsError@users.noreply.github.com>
π¨ CVE-2026-70620
Odysseus before commit 87babb5 contains a server-side request forgery vulnerability that allows admin-privileged attackers to direct the server to probe internal network resources by supplying arbitrary URLs to the embedding endpoint configuration without scheme, host, IP range, or DNS rebind validation. Attackers can submit loopback addresses, RFC 1918 ranges, or link-local addresses through the embedding endpoint API to partially read responses from cloud instance metadata services, internal APIs, and other hosts reachable from the server.
π@cveNotify
Odysseus before commit 87babb5 contains a server-side request forgery vulnerability that allows admin-privileged attackers to direct the server to probe internal network resources by supplying arbitrary URLs to the embedding endpoint configuration without scheme, host, IP range, or DNS rebind validation. Attackers can submit loopback addresses, RFC 1918 ranges, or link-local addresses through the embedding endpoint API to partially read responses from cloud instance metadata services, internal APIs, and other hosts reachable from the server.
π@cveNotify
Yunus AydΔ±n Blog
Odysseus: Broken Access Control + SSRF in AI Embedding Endpoint (CVE-2026-70619, CVE-2026-70620)
Odysseus CVE-2026-70619/CVE-2026-70620: broken access control + SSRF. POST /api/embeddings/endpoint has no admin check. Hijack embedding URL for SSRF and exfiltration of chat, RAG and vault data.