EUVD-2026-103522
GHSA-x953-qc42-3hqh
Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain
-
AccessLure victim to malicious RTMP server
-
DeliveryVictim client connects and negotiates
-
ExploitSend oversized AMF3 string length
-
ExecutionAMF3_Decode skips failure check
-
PersistOOB read leaks heap / DecodeTEA writes OOB
-
ImpactHeap corruption and memory disclosure
Vulnerability Assessment
| Weakness |
CWE
ASVS 5.0
V15 - Secure Coding and Architecture Bounds checking, write limits
MITRE ATT&CK
2 techniques across 2 tactics: Execution, Privilege Escalation The technique list with ATT&CK links and the kill-chain view is part of the Basic plan and up. |
| Exploitation Conditions | Exploitation requires the victim to run librtmp/rtmpdump as a client and connect to an attacker-controlled RTMP server, OR an attacker in a man-in-the-middle position on the RTMP connection - this is the passive user interaction (UI:P) and attack requirement (AT:P) in the vector. The heap out-of-bounds WRITE specifically requires the client to be invoked with the --token option (which triggers the DecodeTEA code path), a non-default configuration; without it, the realistic impact is limited to the out-of-bounds READ / heap byte leak. Limiting factors: no victim-side exposure unless an RTMP client is actively used, RTMP is a legacy/declining protocol, and the most damaging write primitive depends on the non-default token option being enabled. |
| Impact | This is a moderate-priority issue rather than a top-tier emergency. The input CVSS 4.0 vector (CVSS:3.1-equivalent AV:N) indicates a network attack vector, but the attack requirement (AT:P) and passive user interaction (UI:P) mean this is not a server-side flaw an attacker can hit on demand - the victim must actively run an RTMP client (rtmpdump/librtmp) and connect to attacker-controlled or MITM'd infrastructure. Impact is split: an out-of-bounds read leaks heap bytes (VC:L/C:L) in the baseline case, while the more serious heap out-of-bounds write (reflected in VA:H / availability) is gated behind use of the --token option invoking DecodeTEA, a non-default configuration. No EPSS score was supplied in the input and the CVE is NOT in CISA KEV, so there is no confirmed active exploitation. No public exploit code is identified in the input, though a detailed technical write-up (hackmd) exists that lowers the barrier to developing one. In short: a real but conditional memory-corruption bug in a legacy media library, not a mass-exploitable critical. EPSS is absent (not low - simply not provided), so widespread-vs-targeted exploitation cannot be inferred from that signal. |
| Exploit ScenarioDerived | A user runs rtmpdump/librtmp to fetch a stream from a malicious or compromised RTMP server (or an attacker on-path performs a man-in-the-middle). The server returns an AMF3 object containing an oversized AMF3 string length; because AMF3_Decode ignores property decode failures, librtmp performs an out-of-bounds read that leaks heap memory, and when the client was invoked with the --token (DecodeTEA) option, the malformed length drives a heap out-of-bounds write corrupting heap memory. No public exploit code is identified at time of analysis, though a public technical write-up describes the flaw in detail. |
Get an e-mail when this CVE is exploited, patched or re-scored. Free account.
Alert meAnalysisAIv1 – Oct 18, 2026
Out-of-bounds heap reads, and under a non-default configuration out-of-bounds heap writes, in librtmp as shipped with RTMPDump through 2.6 let an unauthenticated attacker who controls or intercepts an RTMP server corrupt or leak memory inside the connecting client. AMF3_Decode ignores failures returned by the AMF3 property decoder, so an oversized AMF3 string length supplied by the server is trusted and used against buffers it does not actually describe, producing an over-read of heap bytes; the write primitive is only reached when the client is invoked with the --token option that enables the DecodeTEA path. Scored 6.1, with no public exploit identified at time of analysis, and real exposure confined to hosts that actively act as RTMP clients against attacker-reachable servers.
RemediationAI
No vendor-released patch identified at time of analysis: RTMPDump 2.6 is the latest release referenced and the provided intelligence contains no fixed version, so the reference to the upstream amf.c blob and the VulnCheck advisory should be tracked until a corrected librtmp build appears, after which any application statically linking the library must be rebuilt. Until then, the highest-value control is to remove the heap-write primitive by dropping the --token option (and any equivalent token/DecodeTEA setting) from rtmpdump and librtmp-based pipelines - the trade-off is that servers requiring token authentication will no longer complete the handshake. Restrict outbound RTMP (TCP 1935, plus RTMP tunnelled over 443/80) to an allowlist of known-good streaming origins at the egress firewall or proxy, accepting that legitimate streams to unlisted CDNs will break. Do not run librtmp clients against untrusted, user-supplied or MITM-exposed RTMP URLs, and where server-side RTMP ingest is unavoidable, terminate it in a maintained implementation or an AMF3-validating proxy rather than librtmp. Monitor hosts running rtmpdump/FFmpeg librtmp for abnormal crashes or memory growth that could indicate a leaked-heap or corrupted-heap condition, and plan migration off librtmp given RTMP's declining use and the absence of a maintained upstream fix.
Affected VersionsENISA
Affected Packages
Source: OSV (Open Source Vulnerabilities)
Affects your stack
Affects 1 package in euvd
Checking this CVE against your own stacks and servers is part of the Basic plan and up.
Technical ContextAI · show details
librtmp implements the client side of RTMP and its AMF0/AMF3 serialization layer in librtmp/amf.c; AMF3 strings are length-prefixed with a variable-length U29 encoding, and AMF3_Decode is expected to validate that prefix against the bytes actually remaining in the received buffer before copying. The defect is that AMF3_Decode discards the status returned by the nested AMF3 property decoder, so when a decode fails (for example because a hostile server advertises a string length far larger than the message body), execution continues with an unvalidated length and pointer, which is the classic unchecked-length condition underlying CWE-787 (out-of-bounds write) and the paired out-of-bounds read. The write side additionally depends on the token-authenticated handshake path, where DecodeTEA decrypts server-supplied data with the Tiny Encryption Algorithm and then hands it to the AMF3 parser - without strict length checks, attacker-influenced ciphertext drives heap writes past the allocation while the same path leaks adjacent heap bytes back through parse results. Because librtmp is a shared library rather than a standalone service, the vulnerable parser is inherited by every consumer that links it, including rtmpdump, FFmpeg builds configured with --enable-librtmp, and third-party streamers and downloaders.
Know when Librtmp gets its next vulnerability
- Watch products and get an email when something new is published
- Check this CVE against your own stack and servers
- Exploit and PoC references, threat intelligence from 22 sources
No credit card · 7-day full trial
Severity by source
CVSS:4.0/AV:N/AC:L/AT:P/PR:N/UI:P/VC:L/VI:L/VA:H/SC:N/SI:N/SA:N
Remote malicious/MITM server (AV:N, PR:N) but victim must run the client (UI:R); heap leak gives C:L and the token-gated OOB write yields corruption/crash (A:H).
Primary rating from Vendor (VulnCheck).
CVSS VectorVendor: VulnCheck
DescriptionCVE.org
librtmp in RTMPDump through 2.6 contains an out-of-bounds read and heap out-of-bounds write vulnerability because AMF3_Decode ignores AMF3 property decode failures. Malicious RTMP servers or man-in-the-middle attackers can send oversized AMF3 string lengths, and with the token option DecodeTEA corrupts heap memory and leaks heap bytes.
Threat Intelligence Overview
NVD
CVE.org
VulDB
How this decision was reached
The engine reads each signal below and applies a fixed list of rules from top to bottom. The first rule that matches sets the recommended action – like a statement, every line is shown.
What every action level means
SSVC AssessmentENISA
Priority Score
Attack Technique
A buffer overflow occurs when a program writes more data to a memory buffer than it was allocated to hold, causing the excess data to spill into adjacent memory regions.
Memory corruption occurs when a program writes data beyond the boundaries of allocated memory regions or accesses memory in unintended ways, violating the integrity of the process's address space.
Exploit-DB