Intelligence
Explorer.

Discourse is an open-source discussion platform. Prior to 2026.1.8, 2026.6.3, 2026.7.2, and 2026.8.0, the Discourse post sanitizer allowed a stored cross-origin iframe to bypass the allowed_iframes prefix policy when the iframe src contained encoded userinfo. The sanitizer validated a decoded form differently from the stored iframe src, allowing the browser to interpret an attacker-controlled host while the allowlist check accepted the encoded URL as an allowed prefix. An authenticated user with posting privileges could persist the iframe in a post and cause attacker-controlled cross-origin content to be rendered. This issue is fixed in versions 2026.1.8, 2026.6.3, 2026.7.2, and 2026.8.0.

Discourse is an open-source discussion platform. Prior to 2026.1.8, 2026.6.3, 2026.7.2, and 2026.8.0, authenticated users could supply unescaped SQL LIKE metacharacters to upload-resolution patterns, causing wildcard input to select unrelated upload records instead of matching a literal identifier. The affected upload metadata, URL-lookup, and cooked-video-placeholder paths could resolve uploads the user was not authorized to access. This allowed disclosure of restricted metadata, including original filenames and secure-upload paths, without modifying the upload records. This issue is fixed in versions 2026.1.8, 2026.6.3, 2026.7.2, and 2026.8.0.

Discourse is an open-source discussion platform. Prior to 2026.1.8, 2026.6.3, 2026.7.2, and 2026.8.0, sites using wildcard patterns in the allowed_iframes setting could accept a crafted iframe URL whose allowlisted suffix appeared after a URL authority separator. The wildcard origin check matched the allowed domain text, while browser URL parsing selected a different attacker-controlled origin. A user with posting privileges could exploit the flaw through a post or a Onebox oEmbed response and cause an attacker-controlled iframe to be rendered independently of the allowlisted domain. Successful exploitation requires a site-configured wildcard iframe pattern and the ability to create a post or trigger an oEmbed request for an attacker-controlled URL. This issue is fixed in versions 2026.1.8, 2026.6.3, 2026.7.2, and 2026.8.0.

Discourse is an open-source discussion platform. Prior to 2026.1.8, 2026.6.3, 2026.7.2, and 2026.8.0, the iframe src traversal guard did not treat literal backslashes as path separators after decoded dot segments. A crafted source could therefore pass an allowed_iframes subpath check while browser URL normalization moved the iframe outside the intended allowed path. The resulting iframe could load content from a location that the administrator did not allow. This issue is fixed in versions 2026.1.8, 2026.6.3, 2026.7.2, and 2026.8.0.

Discourse is an open-source discussion platform. Prior to 2026.1.8, 2026.6.3, 2026.7.2, and 2026.8.0, the video placeholder component allowed crafted HTML to cause an attribute breakout and inject an attacker-controlled event handler. An authenticated user with default trust-level posting privileges could store the crafted placeholder in a post. When another user opened the post and clicked the video play overlay, the handler could execute arbitrary JavaScript in the viewer's session. Default Content Security Policy settings block inline event handlers, but instances with CSP disabled or relaxed could allow the script to read page content and make authenticated requests as the viewer. This issue is fixed in versions 2026.1.8, 2026.6.3, 2026.7.2, and 2026.8.0.

An out-of-bounds write vulnerability in jslGetTokenValueAsString() in Espruino 2v29 (commit bffc6d0) allows crafted JavaScript input containing an overlong token to trigger a one-byte write beyond the JsLex.token buffer in RELEASE/NO_ASSERT builds. The out-of-bounds write corrupts the adjacent tokenValue pointer, resulting in memory corruption and potentially causing application crashes or denial of service.

Mini-XML 4.0.5 contains a memory leak vulnerability in mxml_load_data() during malformed XML parsing. Specially crafted XML input can cause text nodes allocated by mxmlNewText() to become unlinked before a parse error transfers control to the cleanup path. These orphaned nodes are not released, resulting in a persistent memory leak on each parsing attempt. Repeated attacker-controlled requests can cause cumulative memory exhaustion and denial of service.

OpenEXR 3.4.14 contains a NULL Pointer Dereference in the C++ attribute parsing path. A specially crafted EXR file containing an unknown-type attribute with dataSize set to zero causes the parser to create an opaque attribute with a NULL packed_data pointer. The OpaqueAttribute constructor passes the NULL pointer to memcpy() without validating the zero-size condition, resulting in undefined behavior and process termination, leading to denial of service.

libical 4.0.6 contains an incompatible function pointer in icalparameter_string_to_kind(). When parsing iCalendar data containing a parameterized property, the function passes icalparameter_compare_kind_map() to bsearch() through an incompatible comparator function pointer type. bsearch() invokes the callback through the mismatched type, resulting in undefined behavior and process termination, leading to denial of service.

hiredis commit 29ea279 (post-v1.5.0) contains an uncontrolled memory allocation vulnerability in its RESP aggregate parser.

QuickJS commit 04be24600 contains a heap out-of-bounds write condition in JS_ReadFunctionTag().

Bento4 1.6.0.0 contains an integer underflow vulnerability in the avcC and hvcC configuration atom parsers. A specially crafted MP4 file containing an atom with a declared size smaller than AP4_ATOM_HEADER_SIZE can cause AP4_AvccAtom::Create() or AP4_HvccAtom::Create() to underflow the payload-size calculation. The resulting oversized buffer operation can cause invalid or NULL pointers to be passed to the AP4_DataBuffer copy path, resulting in application termination and denial of service.

Bento4 1.6.0.0 contains an integer underflow vulnerability in AP4_AvccAtom::Create() and AP4_HvccAtom::Create(). A specially crafted MP4 file containing an avcC or hvcC atom with a declared size smaller than the atom header size can cause the payload-size calculation to wrap to a large unsigned value. The resulting invalid buffer allocation and copy operations can cause application termination, leading to denial of service.

libde265 commit 4d45a6b contains a NULL pointer dereference vulnerability in the NAL parsing path. When de265_push_NAL() is called with a zero-length NAL unit, the resulting NAL_unit may retain a NULL backing buffer, which is subsequently passed as the destination argument to memcpy() in NAL_unit::set_data(). Although the copy length is zero, this violates the nonnull requirement of memcpy() and results in undefined behavior, causing process termination in UBSan-instrumented builds and denial of service.

libsndfile 1.2.2 contains an integer overflow vulnerability in mat4_read_header() when parsing crafted MAT4 (MATLAB v4) files.

NanoSVG 239e102ec contains an incorrect numeric conversion vulnerability in nsvg__curveDivs() during SVG stroke rasterization. A specially crafted SVG document containing an extremely large stroke-width can cause floating-point rounding to produce a zero subdivision angle. The subsequent arc division yields infinity, which is converted to int without range validation, resulting in undefined behavior and process termination, leading to denial of service.

Discourse is an open-source discussion platform. Prior to 2026.1.6, 2026.5.2, 2026.6.1, and 2026.7.0, Discourse AI reviewables associated with private messages could appear in the moderator review queue of a moderator who was not a participant in the message. Reviewable visibility filtering did not restrict private-message reviewables to the audience permitted to access the underlying private-message topic, allowing the moderator to read otherwise confidential content. Depending on the available reviewable action, the moderator could also modify the private message by closing its topic or deleting a post. Exploitation requires an authenticated moderator account and a pre-existing Discourse AI reviewable associated with a private message. This issue is fixed in versions 2026.1.6, 2026.5.2, 2026.6.1, and 2026.7.0.

Termix is a web-based server management platform with SSH terminal, tunneling, and file editing capabilities. From 2.4.1 until 2.5.1, an authenticated Termix administrator can store attacker-controlled domain and email values through PATCH /users/acme-ssl-settings and trigger their interpolation into a certbot shell command through POST /users/acme-ssl-request. In src/backend/database/routes/acme-ssl-routes.ts, child_process.execSync invokes /bin/sh -c with those values only wrapped in double quotes, so shell metacharacters can execute arbitrary operating-system commands as the Termix backend process. Both HTTP webroot and DNS Cloudflare challenge modes are affected, and compromise exposes Termix databases, process secrets, stored credentials, and network reachability. This issue is fixed in version 2.5.1.

Termix is a web-based server management platform with SSH terminal, tunneling, and file editing capabilities. From 2.5.0 until 2.5.1, the /homepage/proxy endpoint accepts an authenticated user's url query parameter and passes it to http.get or https.get without destination restrictions. In src/backend/database/routes/homepage-proxy-routes.ts, new URL performs only syntactic validation, allowing requests to loopback, RFC1918, link-local, and cloud metadata destinations. The endpoint returns the complete fetched JSON response, so a low-privilege or self-registered account can exfiltrate internal service data and cloud credentials. This issue is fixed in version 2.5.1.

Termix is a web-based server management platform with SSH terminal, tunneling, and file editing capabilities. From 2.4.0 until 2.5.1, the POST /users/totp/disable and POST /users/totp/backup-codes endpoints accept the account password as the sole reauthentication factor after a 2.4.0 refactor regressed the two-factor check introduced for CVE-2026-45749. In src/backend/database/routes/user-totp-routes.ts, verifyTotpReauth returns success when bcrypt.compare validates the password, while each endpoint chooses password or totp_code as an interchangeable credential. An attacker who has a victim's authenticated session and knows the password can disable TOTP or regenerate and invalidate backup codes without an authenticator or valid second factor, weakening the account to single-factor authentication. This issue is fixed in version 2.5.1.

Termix is a web-based server management platform with SSH terminal, tunneling, and file editing capabilities. From 1.7.0 until 2.5.1, Termix derives the keys that wrap OIDC and WebAuthn users' Data Encryption Keys from committed default strings and the public userId salt in src/backend/utils/user-crypto.ts. Because OIDC_SYSTEM_SECRET and WEBAUTHN_SYSTEM_SECRET are not configured by the project's default deployment artifacts, an attacker with an offline SQLite database copy can derive the wrapping key, recover each affected user's DEK, and decrypt stored SSH passwords, private keys, and key passphrases. Password-authenticated users are not affected by this specific key derivation path. This issue is fixed in version 2.5.1.

Snipe-IT is an IT asset/license management system. Prior to 8.7.0, the uploaded-files API endpoint GET /api/v1/{object_type}/{id}/files/{file_id} allows an authenticated user with file-management access to upload XML and XSLT attachments and request them with the inline=true parameter. The app/Http/Controllers/Api/UploadedFilesController.php show() path does not apply the safe-inline allowlist used by the equivalent web controller, so the browser can process an attacker-controlled xml-stylesheet reference and execute JavaScript generated by the stylesheet in the Snipe-IT origin. A victim who is authorized to view the object must open the attachment URL, after which the script can read same-origin data and perform authenticated actions with the victim's privileges. This issue is fixed in version 8.7.0.

Snipe-IT is an IT asset/license management system. Prior to 8.7.0, a password-authenticated session for an account with self.api permission can reach the personal-access-token API flow before completing the account's second-factor challenge because CheckForTwoFactor is enforced in the web middleware group but not the API middleware group. The advisory states that the resulting persistent API token can read and modify resources with the victim's permissions and, for an administrator, can reach the users/two_factor_reset endpoint. Resetting the administrator's enrolled second factor allows the password-holding attacker to enroll an attacker-controlled factor, take over the administrator's web account, and lock out the legitimate user. The token does not create a web session, but it provides broad API access while the same browser session remains blocked at the two-factor page. This vulnerability is fixed in 8.7.0.

Snipe-IT is an IT asset/license management system. Prior to 8.7.0, a user with the customfields.create permission can store markup in CustomField.name, and app/Presenters/AssetPresenter.php assigns that value as an unescaped bootstrap-table header title. When another user opens an asset-list page associated with the fieldset, the stored markup executes on page load in that user's Snipe-IT session. This can expose same-origin data and perform authenticated actions with the victim's privileges, including privilege escalation when a superuser views the affected list. This issue is fixed in version 8.7.0.