🚨 CVE-2026-100844
MONAI before 1.6.0 is vulnerable to OS command injection in the nnUNetV2Runner component (monai.apps.nnunet.nnunetv2_runner). User-controlled values taken from the YAML configuration file (notably dataset_name_or_id) and from CLI/kwargs arguments are concatenated into a command string without quoting or validation and then passed to subprocess with shell=True, so shell metacharacters (e.g., ';' on Linux, '&' on Windows) are interpreted. If a victim loads and processes a crafted configuration file — for example by instantiating nnUNetV2Runner with the malicious YAML and invoking a training/validation job such as train_single_model() — arbitrary commands are executed with the privileges of the user running the job.

🎖@cveNotify

🚨 CVE-2026-100845
MONAI before 1.6.0 contains an unsafe deserialization vulnerability in the NumpyReader class that unconditionally uses numpy.load with allow_pickle=True when loading .npy and .npz files. Attackers can craft malicious .npy files with pickle payloads that execute arbitrary code when loaded through MONAI's standard data pipeline.

🎖@cveNotify

🚨 CVE-2026-100846
MONAI before 1.5.2 contains a deserialization of untrusted data vulnerability in the algo_from_pickle function in monai/auto3dseg/utils.py. The function reads a .pkl file and passes its contents to pickle.loads without validating the data source or content. If an application invokes algo_from_pickle on an attacker-supplied pickle file, an object defining __reduce__ is executed during deserialization, resulting in arbitrary code execution in the context of the application.

🎖@cveNotify

🚨 CVE-2026-100847
AzuraCast before 0.23.8 contains a DQL injection vulnerability in the sortOrder API parameter of AbstractSearchableListAction.php. Attackers can inject arbitrary DQL expressions through the sortOrder parameter to extract sensitive database information including user credentials and station settings.

🎖@cveNotify

🚨 CVE-2026-100848
AzuraCast (Composer package azuracast/azuracast) before 0.23.8 validates a station's "Remote Relay" URL only for URL syntax and an http/https scheme (Utilities\Urls::parseUserUrl, used by StationRemote::getUrlAsUri) and performs no host or IP address restriction. A user holding only the station-scoped RemoteRelays permission can therefore set a Remote Relay URL pointing at loopback, private-network, or cloud-metadata addresses (e.g. http://127.0.0.1:<port>/ or http://169.254.169.254/latest/meta-data/), and AzuraCast's periodic background Now Playing sync (AbstractRemote::getNowPlayingAsync) will automatically and repeatedly issue HTTP requests to that address, resulting in server-side request forgery against internal resources. This affects the main branch as of commit bcf8754eef3a268ad82c3db4d81f7920c3c28b56 (2026-07-31); no patched version is available at the time of the advisory.

🎖@cveNotify

🚨 CVE-2026-100849
AzuraCast is a self-hosted web radio management suite. In AzuraCast before 0.23.8, the station webhook URL validation in AbstractConnector::getValidUrl() (backend/src/Webhook/Connector/AbstractConnector.php), used by the Generic and Discord webhook connectors, rejects only URLs whose host is a literal link-local IP address (169.254.0.0/16 or fe80::/10). Loopback addresses and RFC1918 private ranges are not rejected, and any non-literal-IP hostname causes the IP parsing call to throw, which skips the check entirely. A user holding only the station-scoped WebHooks permission can therefore configure a webhook pointing at an internal, loopback, or private-network target and cause the server to issue an outbound HTTP POST containing the station's Now Playing data, resulting in server-side request forgery. The PUT /station/{id}/webhook/{id}/test endpoint allows the same low-privileged user to trigger the request on demand. At the time of the advisory no patched version was available.

🎖@cveNotify

🚨 CVE-2026-100850
AzuraCast before 0.23.8 contains a server-side request forgery and local file read vulnerability in the AutoDJ remote playlist fetch (backend/src/Radio/AutoDJ/QueueBuilder.php, getMediaFromRemoteUrl()). A user with the station Media permission can create or update a playlist with source=remote_url and remote_type=playlist whose remote_url points at a file:// path or an internal/loopback/link-local HTTP endpoint. When AutoDJ builds the queue, the backend passes the user-supplied URL directly to file_get_contents() with no scheme allowlist and no private/loopback/metadata IP policy (PHP allow_url_fopen is enabled by default, including in the Docker image). Lines from the fetched resource are parsed as M3U/PLS entries, stored in StationQueue.autodj_custom_uri, and returned by GET /api/station/{station_id}/queue to any user with the Broadcasting permission, disclosing host files readable by the web container (for example /etc/passwd or the application .env) and the bodies of non-blind internal HTTP requests. No patched version was available at the time of publication.

🎖@cveNotify

🚨 CVE-2026-100851
AzuraCast before 0.23.8 contains a broken access control vulnerability in the GET /api/station/{id}/vue/profile endpoint that allows authenticated users with only View Station Page permission to read Icecast/Shoutcast admin, source, and relay passwords. Attackers with View-only access can call this endpoint and receive plaintext frontend credentials in the JSON response, then use the admin password to authenticate to the Icecast admin interface without Broadcasting permission.

🎖@cveNotify

🚨 CVE-2026-100852
AzuraCast through 0.23.x contains a command injection vulnerability in the Liquidsoap config generation for live recording that fails to quote the streamer username in process.run calls. Authenticated station users with Streamers and Profile permissions can set a username containing shell metacharacters and trigger command execution as the Liquidsoap process user when recording closes.

🎖@cveNotify

🚨 CVE-2026-100853
In AzuraCast before 0.23.8, the public On-Demand download endpoint fails to verify playlist-level access controls, allowing unauthenticated users to download media files excluded from On-Demand-enabled playlists. Attackers can bypass the station operator's intended access restrictions by directly requesting media via the download endpoint using valid media identifiers, exposing private or restricted audio content.

🎖@cveNotify

🚨 CVE-2026-100854
AzuraCast before 0.23.6 lacks RequireInternalConnection middleware on the Liquidsoap API endpoint and incorrectly derives the AutoDJ flag from header presence rather than validated value. Users with View station permission can inject arbitrary now-playing metadata, disrupt live broadcasts, and disclose filesystem paths.

🎖@cveNotify

🚨 CVE-2026-100855
AzuraCast before 0.23.6 contains a missing permission check vulnerability in the GET /api/station/{station_id}/file/{id}/play endpoint that allows authenticated users to download media files from any station. Attackers can enumerate media files using sequential IDs and exfiltrate the complete media library of stations they lack permissions for.

🎖@cveNotify

🚨 CVE-2026-100856
AzuraCast before 0.23.6 contains a code injection vulnerability in the remote relay password field due to incomplete migration from the vulnerable cleanUpString method to toRawString. Attackers with RemoteRelays station permission can inject nested Liquidsoap interpolation syntax to execute arbitrary code in the Liquidsoap process, disclose internal API keys, or disrupt station operation.

🎖@cveNotify

🚨 CVE-2026-100857
AzuraCast before 0.23.4 contains a code injection vulnerability in the ConfigWriter::cleanUpString() method that fails to sanitize Liquidsoap string interpolation sequences, allowing authenticated users with Media or Profile permissions to inject arbitrary Liquidsoap code into station configuration. Attackers can inject #{process.run()} expressions into playlist URLs or station metadata fields that execute shell commands as the azuracast user when the station restarts.

🎖@cveNotify

🚨 CVE-2026-100858
heym before 0.0.109 contains a server-side request forgery vulnerability in the Slack, Discord, and Crawler workflow nodes. These nodes issue HTTP requests to URLs taken from user-created credentials (webhook_url / flaresolverr_url) using an unguarded HTTP client, bypassing the SSRF egress guard that already protects the HTTP, WebSocket, and MCP nodes; the credential API validates only that the URL is non-empty. Any registered user can create a credential pointing at an internal address and execute a workflow, causing the backend to reach loopback, private, link-local, or cloud-metadata endpoints and return the full response body in the node output (non-blind SSRF).

🎖@cveNotify

🚨 CVE-2026-100859
Heym before 0.0.106 contains a credential exfiltration vulnerability in the POST /api/credentials/test endpoint that allows collaborators with shared credential access to exfiltrate the credential owner's secret. Attackers can override the destination URL in the config parameter to cause the server to send decrypted authentication secrets to attacker-controlled endpoints.

🎖@cveNotify

🚨 CVE-2026-100860
heym before 0.0.105 does not act on the result of the credential authorization lookup in the Redis workflow node (backend/app/services/node_execution/nodes/redis_node.py). When _get_accessible_credential returns None — because the credential ID does not exist or the caller is not authorized to use it — the node treats the lookup failure as an empty configuration and falls back to defaults, connecting to localhost:6379 with no password and executing the requested operation there. The same fallback occurs when an accessible credential has an empty config or no redis_host value. An authenticated workflow author who supplies a credential ID they do not own, or one that was deleted, therefore obtains a read/write connection to whatever Redis is listening on the backend's loopback interface instead of an error. Impact depends on the deployment: the stock docker-compose.yml ships no Redis, in which case the flaw surfaces as a misleading connection error rather than data exposure.

🎖@cveNotify

🚨 CVE-2026-100861
heym before 0.0.105 fails to apply egress guards to integration services that use credential-supplied base URLs, allowing authenticated users to bypass SSRF protections. Attackers can configure credentials pointing to loopback, private, or cloud-metadata addresses and read internal service responses returned as workflow node output.

🎖@cveNotify

🚨 CVE-2026-100863
Heym versions 0.0.90 and earlier contain two server-side request forgery (SSRF) egress gaps, both remediated in app/services/ssrf_guard.py in 0.0.91. First, the LLM image-edit input loader (_load_image_bytes) fetched caller-controlled HTTP/HTTPS URLs with a bare httpx.get, applying only a scheme check and bypassing the egress-pinning HTTP client; because the workflow DSL supports "imageInput": "$userInput.body.imageUrl", a webhook or API caller can choose the fetch target when a workflow author uses that expression, allowing requests to loopback, RFC1918, and cloud metadata endpoints. Second, _is_public_address unwrapped only IPv4-mapped IPv6 addresses, so IPv6 transition forms — the NAT64 well-known prefix 64:ff9b::/96, deprecated IPv4-compatible ::x.x.x.x addresses, and 6to4 (2002::/16, classified as globally routable by Python 3.11.0 through 3.11.9) — could carry loopback, RFC1918, link-local, or cloud-metadata IPv4 destinations past both the initial URL validation and the dial-time IP pin. Version 0.0.91 routes the image loader through guard_http_url and the guarded client, evaluates NAT64 and IPv4-compatible addresses by their embedded IPv4 address, and refuses 64:ff9b:1::/48, 6to4, and Teredo (2001::/32) outright.

🎖@cveNotify

🚨 CVE-2026-100864
heym before 0.0.91 contains a sandbox escape vulnerability in the expression engine's DotList map/filter and fallback resolver that allows authenticated users to execute arbitrary Python code. Attackers can craft workflow expressions using dunder attribute access through item expressions or the fallback resolver to access os.system and execute commands as the backend process.

🎖@cveNotify

🚨 CVE-2026-100865
Heym before 0.0.53 contains multiple independent vulnerabilities. (1) The workflow condition evaluator uses Python eval() without an effective sandbox, allowing any user who can edit a workflow branch/condition node — or who can import a workflow template containing a malicious condition node — to execute arbitrary Python code as the backend process user when the workflow runs. (2) Slack webhook signature verification and (3) Telegram webhook secret-token verification fail open when the trigger node has no credentialId or an empty signing secret, allowing anyone who knows the public webhook URL to trigger workflows with the owner's credentials without authentication. (4) The OAuth authorization endpoint does not validate the redirect_uri scheme, so an attacker who registers a public OAuth client with a javascript: or data: redirect_uri and lures a victim to the consent screen receives the authorization code and executes attacker-controlled JavaScript in the Heym origin, including access to the victim's HttpOnly auth cookie. (5) WorkflowExecutionToken, PortalSession, HITLRequest.public_token, and OAuthAuthorizationCode values are stored in plaintext, so any database read exposure yields valid scoped bearer tokens, including portal and HITL tokens with a 168-hour TTL that permit workflow execution on behalf of the owner.

🎖@cveNotify