GitLab Advisory Database (GLAD)

YOURLS stores the HTTP Referer header for short URL redirects and later renders aggregated referrer domains in the per-link statistics page. An unauthenticated attacker can send a crafted Referer header to any existing short URL. When an authenticated administrator or stats-page viewer opens that short URL's statistics page, the crafted referrer is embedded into Google Charts JavaScript without JavaScript-string escaping, causing stored cross-site scripting. This is reachable in default private …

Xinference used Python's unsafe eval() function when parsing Llama3 tool-call output generated by a large language model. Because the model output can be influenced by attacker-controlled prompts sent to the chat completion API, a remote attacker can craft prompts that cause the model to return a Python expression. Xinference then evaluates that expression on the server while post-processing the tool-call result. In the tested default deployment, authentication was not enabled, …

An unauthenticated POST to any OpenAPI-validated endpoint, including the anonymous POST /edge/validate and POST /edge/issue-token, crashes the entire Unleash server with one request body of deeply-nested JSON. When request-body validation fails, Unleash builds the error message by calling JSON.stringify on the raw offending value taken from the request body. A value nested a few thousand levels deep makes JSON.stringify recurse past the V8 call-stack limit and throw RangeError: Maximum call …

Stored markdown/link-injection (phishing-link injection) into any configured outbound notification channel (Slack legacy, MS Teams, Webhook default markdown, Datadog, New Relic), using an attacker-controlled username — no admin privilege required, only Editor on a single project, and potentially reachable through public self-signup. Secondary: loss of HTML escaping for any other reachable Mustache single-mustache placeholder process-wide until restart (increases severity of any other currently-unreached or future Mustache sink, e.g. email templates). Tertiary: …

Unleash's addon/integration subsystem lets an operator configure a webhook (and the Slack, Microsoft Teams, Datadog, and New Relic integrations) with a target url parameter. Whenever a subscribed feature-flag event fires, the Unleash server itself issues an HTTP request to that configured URL. The URL is taken verbatim from the addon's parameters.url and passed straight to the HTTP client (ky) with no validation of the host: there is no allow-list, no …

The Volt template compiler in Phalcon generates the PHP for the join filter by string-concatenating the filter's raw template-literal argument bytes with no escaping. The separator literal is dropped verbatim between two single quotes the compiler emits, and the piped array argument is emitted completely bare. A Volt template whose join arguments are attacker-influenced can therefore break out of the generated join('…') call and inject arbitrary PHP into the compiled …

A nil-pointer dereference in openapi3filter.ConvertErrors lets any unauthenticated client crash a server with a single HTTP request. When an application validates a multipart/form-data request body and renders the resulting validation error through the library-provided ValidationErrorEncoder / ConvertErrors helpers, a malformed scalar form field (e.g. a non-numeric value for an integer property) produces an error shape that convertParseError dereferences without a nil check. The handler goroutine panics, causing a denial of …

An uncontrolled resource consumption vulnerability in openapi3filter lets any unauthenticated client force multi-gigabyte heap allocation with a single, tiny HTTP request. When a spec declares a deepObject-style query parameter whose schema contains an array (a normal, documented pattern), the decoder reconstructs the array by reading the largest attacker-supplied index and allocating one slot for every position from 0 up to that index — before schema validation (including maxItems) ever runs. …

Insufficient clearing of the output buffer in Java-based decompressor implementations in lz4-java 1.10.0 and earlier allows remote attackers to read previous buffer contents via crafted compressed input. In applications where the output buffer is reused without being cleared, this may lead to disclosure of sensitive data. JNI-based implementations are not affected.

Insufficient clearing of the output buffer in Java-based decompressor implementations in lz4-java 1.10.0 and earlier allows remote attackers to read previous buffer contents via crafted compressed input. In applications where the output buffer is reused without being cleared, this may lead to disclosure of sensitive data. JNI-based implementations are not affected.

Insufficient clearing of the output buffer in Java-based decompressor implementations in lz4-java 1.10.0 and earlier allows remote attackers to read previous buffer contents via crafted compressed input. In applications where the output buffer is reused without being cleared, this may lead to disclosure of sensitive data. JNI-based implementations are not affected.

This advisory has been marked as a false positive.

This advisory has been marked as a false positive.

This advisory has been marked as a false positive.

This advisory has been marked as False Positive and has been removed.

After an API Token exposure from an exploited trivy dependency, two new releases of litellm were uploaded to PyPI containing automatically activated malware, harvesting sensitive credentials and files, and exfiltrating to a remote API. Anyone who has installed and run the project should assume any credentials available to litellm environment may have been exposed, and revoke/rotate thema ccordingly.