MCPpedia picked up 2 high-severity advisories affecting 1 catalogued MCP server in the last seven days. Details, and what to do about each, below.
CVE-2026-114054 โ HIGH
PYSEC-2026-4362
Affects Trudi.
vLLM before 0.31.0 contains a cache aliasing vulnerability that allows authenticated attackers to collide shared cache keys by supplying caller-controlled values such as EXIF ImageID or cache_salt. Attackers can set a UUID EXIF ImageID to receive another request's cached processed media, or match cache_salt to a LoRA name to reuse first-block KV state.
Published 2026-10-21.
CVE-2026-114052 โ HIGH
PYSEC-2026-4361
Affects Trudi.
vLLM after 0.24.0 and before 0.31.0 contains a memory exhaustion vulnerability in MOSS-Audio model serving that caches a processor for every distinct request-controlled mm_processor_kwargs value in an unbounded dictionary. Unauthenticated attackers can send distinct or oversized mel_config values to the /tokenize endpoint, which bypasses --api-key, to exhaust host memory and crash the API process.
Published 2026-10-21.
What to do
- Check which of these servers your client config actually loads โ an MCP server you installed once and forgot still runs with the permissions you gave it.
- Upgrade to a patched version where the advisory names one.
- Where no patch exists, remove the server from your config until one does.
Each affected server's page on MCPpedia carries its full advisory list and current security score.
Compiled automatically from the MCPpedia catalog on 2026-11-12. Scores are the MCPpedia 0โ100 rating โ security, maintenance, efficiency, documentation and compatibility, weighted as described in the scoring methodology. Browse the full catalog at mcppedia.org/servers, or compare any two servers side by side with the comparison tool.
MCP Security Weekly
Weekly CVE alerts, new server roundups, and MCP ecosystem insights. Free.
Keep reading
This article was written by AI, powered by Claude and real-time MCPpedia data. All facts and figures are sourced from our database - but AI can make mistakes. If something looks off, let us know.