Search vulnerabilities

Results 9,193

Export CSV
CVE-2026-53426

Atom-table exhaustion denial-of-service via JSON parse_document in MDEx

Allocation of Resources Without Limits or Throttling vulnerability in leandrocp MDEx allows Excessive Allocation. MDEx.parse_document/2 accepts a {:json, json} source. In lib/mdex.ex, the private json_to_node/1 function passes the attacker-controlled node_type value to Module.concat/1, which calls String.to_atom/1 and interns a brand-new atom for every distinct value. Atoms are never garbage collected on the BEAM, so a crafted JSON document carrying a unique node_type at each (deeply nested) node mints one permanent atom per node. A single document can intern hundreds of thousands of atoms, and a large enough document exhausts the default atom table (around 1,048,576 atoms) and aborts the entire Erlang VM, taking down every process on the node. Any application that passes untrusted input to the {:json, ...} source of MDEx.parse_document is exposed to an unauthenticated denial-of-service. This issue affects mdex: from 0.4.3 before 0.13.2.

CVSS 4.0 · 8.2 High EPSS < 1% KEV · Not listed
Vendors
Products
CVE-2026-133626

FlowIntel: Resource Exhaustion via Sparse CSV/JSON Table Expansion During File-to-Note Conversion

FlowIntel contains a resource exhaustion vulnerability in its file-to-note conversion feature. When a user converts an attached CSV or JSON file into a case note, the application renders the data as a Markdown table by padding every row to the maximum column count. A crafted file with a small number of filled cells but a very large number of columns (or keys) causes the application to allocate a quadratic number of padded table cells, potentially exhausting a worker process in memory or CPU. The same issue affects nested JSON values rendered inside table cells, where pretty-printed indentation further inflates the output size. Preconditions: - An authenticated user with access to a case in FlowIntel - Ability to attach and convert a CSV or JSON file to a note Impact: - Denial of service: exhaustion of a worker process, potentially disrupting service for other users - No confidentiality or integrity impact Affected versions: FlowIntel prior to 4.1.0

CVSS 4.0 · 7.1 High EPSS < 1% KEV · Not listed
Vendors
Products
CVE-2026-53423

Unauthenticated denial-of-service via BEAM atom table exhaustion in membrane_mp4_plugin

Allocation of Resources Without Limits or Throttling vulnerability in membraneframework membrane_mp4_plugin allows unauthenticated denial-of-service via BEAM atom table exhaustion. The MP4 box header parser converts each 4-byte box name to an atom using String.to_atom/1 without validation. Membrane.MP4.Container.Header.parse_box_name/1 in lib/membrane_mp4/container/header.ex interns every box name encountered while Membrane.MP4.Container.Header.parse/1 walks the input. BEAM atoms are never garbage-collected, so each unique attacker-controlled 4-byte name is a permanent allocation. A crafted MP4 of approximately 8 MB containing roughly 1.1 million boxes with distinct non-standard names exhausts the atom table (default ceiling around 1,048,576 atoms), aborting the entire BEAM node and taking down all applications running on it. This issue affects membrane_mp4_plugin: from 0.3.0 before 0.36.7.

CVSS 4.0 · 5.9 Medium EPSS < 1% KEV · Not listed
CVE-2026-49754

HTTP/2 CONTINUATION flood in Mint client via unbounded header-block accumulation

Allocation of Resources Without Limits or Throttling vulnerability in elixir-mint Mint allows attacker-controlled HTTP/2 servers to exhaust memory in a Mint client (HTTP/2 CONTINUATION flood). When Mint's HTTP/2 receive path observes a HEADERS frame without the END_HEADERS flag, the unparsed header-block fragment is parked in conn.headers_being_processed, and every subsequent CONTINUATION frame on that stream is appended to the accumulator. Nothing in the receive path caps the accumulator: there is no per-stream size limit, no CONTINUATION frame-count limit, and max_header_list_size is only enforced on outgoing requests, never on inbound header blocks (its default is :infinity). A malicious or compromised HTTP/2 server can stream an endless sequence of CONTINUATION frames (each up to the peer-advertised SETTINGS_MAX_FRAME_SIZE) and drive the client's iolist to arbitrary size, causing memory exhaustion and BEAM process death. A single connection to an attacker-controlled HTTP/2 endpoint is sufficient. This issue affects mint: from 0.1.0 before 1.9.0.

CVSS 4.0 · 8.2 High EPSS < 1% KEV · Not listed
Vendors
Products
CVE-2026-48862

Unbounded conn.streams growth in Mint HTTP/2 client via unenforced PUSH_PROMISE concurrency

Allocation of Resources Without Limits or Throttling vulnerability in elixir-mint Mint allows attacker-controlled HTTP/2 servers to exhaust memory in a Mint client via PUSH_PROMISE flooding. In lib/mint/http2.ex, Mint.HTTP2.decode_push_promise_headers_and_add_response/5 inserts a :reserved_remote entry into conn.streams for every promised stream ID. The neighbouring Mint.HTTP2.assert_valid_promised_stream_id/2 only verifies that the promised ID is even and not already present; client_settings.max_concurrent_streams is not consulted at promise time. The concurrency cap is only checked when the response HEADERS for the promised stream arrive, so a server that emits PUSH_PROMISE frames and withholds the matching HEADERS never trips that check. HTTP/2 server push is accepted by default (client_settings.enable_push defaults to true). A single long-lived HTTP/2 connection to a hostile server lets that server pin one conn.streams entry per PUSH_PROMISE frame it sends, with no upper bound, until the client process runs out of memory. This issue affects mint: from 0.2.0 before 1.9.0.

CVSS 4.0 · 8.2 High EPSS < 1% KEV · Not listed
Vendors
Products
CVE-2026-133500

Gladys Assistant Backup Restore Endpoint gateway.downloadBackup.js os command injection

A vulnerability was identified in Gladys Assistant up to 4.71.0. This affects an unknown function of the file gateway.downloadBackup.js of the component Backup Restore Endpoint. The manipulation of the argument file_url leads to os command injection. The attack can be initiated remotely. The exploit is publicly available and might be used. Upgrading to version 4.72.0 mitigates this issue. Upgrading the affected component is recommended.

CVSS 4.0 · 9.4 Critical EPSS 3.0% KEV · Not listed
Products
CVE-2026-10600

Denial of service via unbounded document content extraction in Mattermost Server

Mattermost versions 11.8.x <= 11.8.0, 11.7.x <= 11.7.3, 11.6.x <= 11.6.5, 11.2.x <= 11.2.4, 10.11.x <= 10.11.20 fail to bound the time and resource consumption of server-side document content extraction which allows an authenticated user with file-upload permission to degrade file uploads for all users on the server via repeatedly uploading small documents that are cheap to upload but expensive to extract, saturating the shared extraction worker pool. Mattermost Advisory ID: MMSA-2026-00694

CVSS 3.1 · 4.3 Medium EPSS < 1% KEV · Not listed
Vendors
CVE-2026-47067

Atom table exhaustion via unrecognized URL schemes in hackney

Allocation of Resources Without Limits or Throttling vulnerability in benoitc hackney allows Flooding. The URL parser in src/hackney_url.erl converts every unrecognized URL scheme to a permanent BEAM atom via binary_to_atom/2. BEAM atoms are never garbage-collected and the atom table defaults to a hard limit of 1,048,576 entries. An attacker who can supply URLs with attacker-chosen scheme prefixes — directly as request targets, as configured webhook URLs, or via Location headers followed during redirects — can exhaust the atom table and crash the entire BEAM VM with system_limit. This issue affects hackney: from 2.0.0 before 4.0.1.

CVSS 4.0 · 8.7 High EPSS < 1% KEV · Not listed
Vendors
Products
CVE-2026-4897

Polkit: polkit: denial of service via unbounded input processing through standard input

A flaw was found in polkit. A local user can exploit this by providing a specially crafted, excessively long input to the `polkit-agent-helper-1` setuid binary via standard input (stdin). This unbounded input can lead to an out-of-memory (OOM) condition, resulting in a Denial of Service (DoS) for the system.

CVSS 3.1 · 5.5 Medium EPSS < 1% KEV · Not listed
CVE-2026-43973

gun HTTP/1.1 response buffer has no size limit allowing server-controlled memory exhaustion

Uncontrolled Resource Consumption vulnerability in ninenines gun (gun_http module) allows a malicious server to exhaust client memory via unbounded HTTP/1.1 response buffering. In gun_http:handle/5, three clauses accumulate incoming TCP data into the connection's buffer field using binary concatenation with no upper-bound check: the head clause appends data until the \r\n\r\n header terminator is found; the body_chunked clause appends data whenever cow_http_te:stream_chunked/2 returns a more result indicating an incomplete chunk boundary; and the body_trailer clause appends data until the trailing \r\n\r\n is found. In each case, when the expected terminator never arrives, the enlarged binary is stored back into state and the process waits for more data, with no configurable or hard-coded ceiling on buffer size. A malicious or compromised server can exploit this by sending a partial response that never completes. For example, a response may begin with HTTP/1.1 200 OK\r\nX-Pad: followed by an unbounded stream of arbitrary bytes, never sending the header terminator. The gun connection process will continuously append the incoming data to its buffer, causing unbounded heap growth. Because BEAM imposes no per-process heap limit by default, a single malicious connection can exhaust all available memory on the node, causing a node-wide out-of-memory crash. This issue affects gun: from 1.0.0-pre.1 before 2.4.0.

CVSS 4.0 · 8.7 High EPSS < 1% KEV · Not listed
Vendors
Products
Gun
CVE-2026-8469

Unauthenticated denial-of-service via BEAM atom table exhaustion in phoenix_storybook

Allocation of Resources Without Limits or Throttling vulnerability in phenixdigital phoenix_storybook allows unauthenticated denial of service via atom table exhaustion. LiveView event handlers and request parameters of the storybook convert client-supplied strings to atoms. Atoms are never garbage-collected, so a client that sends enough unique strings fills the atom table, and the BEAM node terminates. PhoenixStorybook.ExtraAssignsHelpers.handle_set_variation_assign/3 and handle_toggle_variation_assign/3 call String.to_atom/1 on parameter keys, attribute names, variation ids, and values of attributes declared as :atom or :boolean. PhoenixStorybook.Story.Playground.handle_event/3 and PhoenixStorybook.StoryLive do the same for tab names, theme names, and playground field names. Releases published as phx_live_storybook have the same calls in the PhxLiveStorybook namespace, at first in PhxLiveStorybook.EntryLive. This issue affects phoenix_storybook: from 0.5.0 before 1.1.0; phx_live_storybook: from 0.2.0 onward.

CVSS 4.0 · 8.2 High EPSS < 1% KEV · Not listed
Vendors
CVE-2026-133488

XiangLi1999 Diffusion-LM infill.py os.system os command injection

A weakness has been identified in XiangLi1999 Diffusion-LM up to 759889d58ef38e2eed41a8c34db8032e072826f4. This vulnerability affects the function os.system of the file improved-diffusion/scripts/infill.py. This manipulation of the argument notes causes os command injection. Remote exploitation of the attack is possible. The exploit has been made available to the public and could be used for attacks. This product follows a rolling release approach for continuous delivery, so version details for affected or updated releases are not provided. The project was informed of the problem early through an issue report but has not responded yet.

CVSS 4.0 · 5.3 Medium EPSS 1.2% KEV · Not listed
Vendors
Products
CVE-2026-8466

Unbounded buffer accumulation in multipart header parsing causes denial of service in cowboy

Allocation of Resources Without Limits or Throttling vulnerability in ninenines cowboy allows denial of service via unbounded buffer accumulation in multipart header parsing. cowboy_req:read_part/3 in src/cowboy_req.erl accumulates incoming request bytes into a Buffer binary with no upper-bound check. When cow_multipart:parse_headers/2 returns more or {more, Buffer2}, the function reads up to Length bytes (default 64 KB) from the request body and recurses with the enlarged buffer. There is no equivalent of the byte_size(Acc) > Length guard present in the sibling function read_part_body/4. An unauthenticated attacker can send a multipart/form-data request whose body never yields a complete header section — for example, a body that never contains the advertised boundary delimiter, or one whose header lines never contain \r\n\r\n — and force the server process to accumulate memory linearly with the bytes the protocol layer is willing to deliver. A handful of concurrent such uploads is sufficient to exhaust BEAM memory. The function has had two names. From 0.10.0 to 2.0.0-pre.3 it is cowboy_req:part/3, reached through cowboy_req:part/1 and cowboy_req:part/2. From 2.0.0-pre.4 it is cowboy_req:read_part/3, reached through cowboy_req:read_part/1 and cowboy_req:read_part/2. Both have the same buffer loop. This issue affects cowboy: from 0.10.0 before 2.15.0.

CVSS 4.0 · 8.2 High EPSS < 1% KEV · Not listed
Vendors
Products
CVE-2026-39803

HTTP/1 chunked body reader ignores length cap in bandit

Allocation of Resources Without Limits or Throttling vulnerability in mtrudel bandit allows unauthenticated remote denial of service via memory exhaustion. The chunked clause of Bandit.HTTP1.Socket.read_data/2 in lib/bandit/http1/socket.ex ignores the caller-supplied :length option when reading HTTP/1 chunked request bodies. Instead of capping the accumulated body at the configured limit (e.g. Plug.Parsers' default 8 MB), do_read_chunked_data!/5 buffers every received chunk into an iolist unconditionally and materializes the entire body as a single binary. The function always returns {:ok, body, ...}, so callers cannot interpose a 413 response. Because Plug.Parsers runs before routing and authentication in the standard Phoenix endpoint, an unauthenticated attacker needs no valid route or credentials. Sending a single Transfer-Encoding: chunked POST request with an arbitrarily large body to any path causes the BEAM process to exhaust available memory and be terminated by the OS OOM killer. The content-length path in the same function correctly enforces the limit and is not affected. This issue affects bandit: from 1.4.0 before 1.11.1.

CVSS 4.0 · 8.7 High EPSS < 1% KEV · Not listed
Vendors
Products
CVE-2026-42793

Atom table exhaustion via attacker-controlled GraphQL SDL names in absinthe

Allocation of Resources Without Limits or Throttling vulnerability in absinthe-graphql absinthe allows unauthenticated denial of service via atom table exhaustion when parsing attacker-controlled GraphQL SDL. Multiple Blueprint.Draft.convert/2 implementations in Absinthe's SDL language modules call String.to_atom/1 on attacker-controlled names from parsed GraphQL SDL documents, including directive names, field names, type names, and argument names. Because atoms are never garbage-collected and the BEAM atom table has a fixed limit (default 1,048,576), each unique name permanently consumes one slot. An attacker can exhaust the atom table by submitting SDL documents containing enough unique names, causing the Erlang VM to abort with system_limit and taking down the entire node. Any application that passes attacker-controlled GraphQL SDL through Absinthe's parser is exposed — for example, a schema-upload endpoint, a federation gateway that ingests remote SDL, or any developer tool that runs the parser over user-supplied documents. This issue affects absinthe: from 1.5.0-alpha.2 before 1.10.2.

CVSS 4.0 · 8.2 High EPSS < 1% KEV · Not listed
Products
CVE-2026-32689

Long-poll NDJSON body splitting causes unbounded memory allocation in Phoenix

Allocation of Resources Without Limits or Throttling vulnerability in phoenixframework phoenix allows a denial of service via the long-poll transport's NDJSON body handling. In Phoenix.Transports.LongPoll.publish/4, when a POST request is received with Content-Type: application/x-ndjson, the request body is split on newline characters using String.split/2 with no limit on the number of resulting segments. An attacker can send a body consisting entirely of newline bytes, causing a 1:1 amplification into a list of empty binaries — a 1 MB body produces approximately one million list elements, an 8 MB body approximately 8.4 million. Each element is then walked by Enum.map, materializing another list of the same size. This exhausts BEAM memory and schedulers, crashing the node and terminating all active sessions. A session token required to reach the vulnerable endpoint is freely obtainable by any client via an unauthenticated GET request to the same URL with a matching Origin header, making this attack effectively unauthenticated. This issue affects phoenix: from 1.7.0 before 1.7.22 and from 1.8.0-rc.0 before 1.8.6.

CVSS 4.0 · 8.7 High EPSS < 1% KEV · Not listed
Products
CVE-2026-39804

WebSocket permessage-deflate inflate has no output-size cap in bandit

Allocation of Resources Without Limits or Throttling vulnerability in mtrudel bandit allows unauthenticated remote denial of service via memory exhaustion when WebSocket permessage-deflate compression is enabled. Bandit.WebSocket.PerMessageDeflate.inflate/2 in lib/bandit/websocket/permessage_deflate.ex calls :zlib.inflate/2 with no output-size cap, then materializes the entire decompressed payload as a single binary via IO.iodata_to_binary/1. The websocket_options.max_frame_size option only bounds the on-the-wire (compressed) frame size, not the decompressed output. A high-ratio compressed frame (e.g. uniform data at ~1024:1 ratio) can stay well under any wire-size limit while forcing GiB-scale heap allocations in the connection process before any application code runs. An unauthenticated attacker who can open a WebSocket connection can send a single such frame to exhaust the BEAM node's memory and trigger an OOM kill. This vulnerability requires both Bandit's server-level websocket_options.compress and the per-upgrade compress: true option passed to WebSockAdapter.upgrade/4 to be enabled. Stock Phoenix and LiveView applications are not affected as they default to compress: false. This issue affects bandit: from 0.5.9 before 1.11.0.

CVSS 4.0 · 8.2 High EPSS < 1% KEV · Not listed
Vendors
Products
CVE-2026-42786

WebSocket fragmented message reassembly unbounded in bandit

Allocation of Resources Without Limits or Throttling vulnerability in mtrudel bandit allows unauthenticated remote denial of service via memory exhaustion. The fragment reassembly path in Bandit.WebSocket.Connection.handle_frame/3 in lib/bandit/websocket/connection.ex appends every incoming Continuation{fin: false} frame's payload to a per-connection iolist with no cumulative size cap. The existing max_frame_size option only bounds individual frames; a peer that streams an unbounded number of continuation frames without ever setting fin=1 grows BEAM heap linearly until the OS or a supervisor kills the process. Because the accumulation happens before WebSock.handle_in/2 is called, the application has no opportunity to interpose a size check. Phoenix Channels and LiveView both run over WebSock on Bandit, so a stock Phoenix application exposes this surface as soon as it accepts socket connections. This issue affects bandit: from 0.5.1 before 1.11.0.

CVSS 4.0 · 8.7 High EPSS < 1% KEV · Not listed
Vendors
Products
CVE-2026-42788

HTTP/2 frame size limit checked after body is buffered in bandit

Allocation of Resources Without Limits or Throttling vulnerability in mtrudel bandit allows unauthenticated memory exhaustion via oversized HTTP/2 frames. Bandit.HTTP2.Frame.deserialize/2 in lib/bandit/http2/frame.ex checks the SETTINGS_MAX_FRAME_SIZE limit only after pattern-matching payload::binary-size(length), which requires the entire frame body to be present in memory before either the accept or reject clause can fire. A peer that announces a frame length up to the 24-bit maximum (~16 MiB) causes the server to buffer that entire body before the size guard is evaluated, regardless of the max_frame_size negotiated during the HTTP/2 handshake (default 16 KiB per RFC 9113). An unauthenticated attacker holding many concurrent connections can force the server to buffer far more memory than the negotiated frame size limit should permit, leading to memory pressure and potential denial of service. This issue affects bandit: from 0.3.6 before 1.11.0.

CVSS 4.0 · 6.9 Medium EPSS < 1% KEV · Not listed
Vendors
Products
CVE-2026-32145

Multipart form body parser bypasses body size limits in wisp

Allocation of Resources Without Limits or Throttling vulnerability in gleam-wisp wisp allows a denial of service via multipart form body parsing. The multipart_body function bypasses configured max_body_size and max_files_size limits. When a multipart boundary is not present in a chunk, the parser takes the MoreRequiredForBody path, which appends the chunk to the output but passes the quota unchanged to the recursive call. Only the final chunk containing the boundary is counted via decrement_quota. The same pattern exists in multipart_headers, where MoreRequiredForHeaders recurses without calling decrement_body_quota. An unauthenticated attacker can exhaust server memory or disk by sending arbitrarily large multipart form submissions in a single HTTP request. This issue affects wisp: from 0.2.0 before 2.2.2.

CVSS 4.0 · 8.7 High EPSS < 1% KEV · Not listed
Vendors
Products