praisonaiagents
Affected
- introduced
0.6.0, fixed1.6.59(ecosystem)
Fixed versions
1.6.59
Fix
- Fixed version
1.6.59pip install --upgrade praisonaiagents==1.6.59
Advisory
Also known as: CVE-2026-57112, GHSA-vmf9-xx9w-86wx, PYSEC-2026-3521
PraisonAI ToolsMCPServer legacy SSE transport accepts attacker Host/Origin and exposes registered tools
PraisonAI ToolsMCPServer legacy SSE transport accepts attacker Host/Origin and exposes registered tools
Summary
praisonaiagents.mcp.ToolsMCPServer.run_sse() builds a Starlette MCP
HTTP+SSE server around mcp.server.sse.SseServerTransport. The server exposes/sse and /messages/, but it does not validate Origin, does not validateHost, and does not require any authentication.
This is reachable through supported PraisonAI code paths that wrap configured
MCP server tools and re-expose them over legacy SSE:
praisonai mcp run <name> --transport ssepraisonai serve mcp --name <name> --transport sseToolsMCPServer(...).run_sse(...) or launch_tools_mcp_server(..., transport="sse")A malicious website can use DNS rebinding against a local or internal
PraisonAI SSE MCP server and send requests with attacker-controlled Host andOrigin headers. The local PoV binds only to 127.0.0.1, sends an attackerHost and Origin, lists the registered tool, and invokes it successfully.
The same attacker Origin is rejected by PraisonAI's current Streamable HTTP
transport with HTTP 403. The vulnerability is therefore a sibling transport
guard gap in the legacy SSE wrapper, not intended behavior.
Affected product
MervinPraison/PraisonAIpraisonaiagentspraisonaisrc/praisonai-agents/praisonaiagents/mcp/mcp_server.pysrc/praisonai/praisonai/cli/commands/mcp.pysrc/praisonai/praisonai/cli/commands/serve.pypraisonaiagents 1.6.58PraisonAI 4.6.581ad58ca02975ff1398efeda694ea2ab78f20cf3eSuggested affected ranges:
praisonaiagents >= 0.6.0, <= 1.6.58praisonai >= 3.10.0, <= 4.6.58No fixed version is known at submission time.
Confirmed source sweep:
v3.0.0 ToolsMCPServer.run_sse helper present, no Origin/Host/auth checks
v3.10.0 praisonai mcp run --transport sse wraps configured tools into helper
v3.12.3 praisonai serve mcp --name --transport sse wraps configured tools
v4.0.0 same vulnerable helper and CLI wrapping paths
v4.4.12 same vulnerable helper and CLI wrapping paths
v4.5.0 same vulnerable helper and CLI wrapping paths
v4.5.56 same vulnerable helper and CLI wrapping paths
v4.5.139 same vulnerable helper and CLI wrapping paths
v4.6.57 same vulnerable helper and CLI wrapping paths
v4.6.58 same vulnerable helper and dynamic PoV succeeds
Impact
If a PraisonAI user starts a local or internal legacy SSE MCP server with
registered tools, an attacker who gets that user to visit a malicious website
can use DNS rebinding to interact with the SSE server through the browser. The
attacker can discover exposed tools and invoke them as the local user.
Impact depends on the configured tools. In realistic PraisonAI MCP deployments,
registered tools may access local files, repositories, issue trackers, cloud
APIs, internal services, or other automation targets. This can lead to
confidentiality, integrity, and availability impact for the resources reachable
by the exposed tools.
The PoV is local-only and harmless. It exposes one marker tool that writes a
canary string to a temporary directory.
Root cause
Current ToolsMCPServer.run_sse() constructs a Starlette app directly:
sse_path = "/sse"
messages_path = "/messages/"
sse_transport = SseServerTransport(messages_path)
async def handle_sse(request: Request):
async with sse_transport.connect_sse(
request.scope, request.receive, request._send
) as (read_stream, write_stream):
await mcp._mcp_server.run(
read_stream,
write_stream,
mcp._mcp_server.create_initialization_options()
)
app = Starlette(
debug=self._debug,
routes=[
Route(sse_path, endpoint=handle_sse),
Mount(messages_path, app=sse_transport.handle_post_message),
]
)
uvicorn.run(app, host=host, port=port)
There is no middleware or route-level check for:
OriginHostAuthorizationThe configured CLI wrapper exposes this path:
from praisonaiagents.mcp import MCP, ToolsMCPServer
cmd_string = " ".join(cmd)
mcp = MCP(cmd_string, timeout=60, env=server.env or {})
tools = mcp.get_tools()
mcp_server = ToolsMCPServer(name=name, tools=tools)
mcp_server.run_sse(host=host, port=port)
By contrast, the current Streamable HTTP transport validates Origin and
returns HTTP 403 for an invalid origin:
origin = request.headers.get("Origin")
if not self._validate_origin(origin):
return JSONResponse(..., status_code=403)
Local-only PoV
Run from the harness checkout:
uv run --with mcp --with starlette --with uvicorn --with httpx --with anyio \
python submission-bundle/praisonai-prai-cand-015-mcp-sse-host-origin-bypass/poc/pov_prai_cand_015_sse_mcp_host_origin_bypass.py \
--repo-src artifacts/repos/praisonai-v4.6.58/src
Observed current-head result:
{
"candidate": "PRAI-CAND-015",
"http_stream_control": {
"attacker_origin": "http://attacker.example.test",
"rejects_attacker_origin": true,
"status_code": 403,
"transport": "current_http_stream"
},
"source_checks": {
"has_auth_check": false,
"has_host_check": false,
"has_origin_check": false,
"has_sse_transport": true,
"route_count": 2
},
"sse_probe": {
"attacker_headers": {
"Host": "attacker.example.test:62380",
"Origin": "http://attacker.example.test:62380"
},
"bind_host": "127.0.0.1",
"marker_value": "executed-from-attacker-origin",
"marker_written": true,
"server_started": true,
"tool_call_content": [
"recorded:executed-from-attacker-origin"
],
"tool_call_error": false,
"tool_names": [
"record_marker"
],
"vulnerable": true
},
"vulnerable": true
}
The PoV:
ToolsMCPServer;uvicorn.run only to capture the exact Starlette app created by run_sse();127.0.0.1;/sse with attacker-controlled Host and Origin;Origin is rejected with HTTP 403.Why this is not intended behavior
This is not only a trust-model disagreement.
PraisonAI's MCP documentation describes Streamable HTTP, WebSocket, and legacy
SSE as supported MCP transport mechanisms. The same documentation says the MCP
module's security properties include origin validation, authentication headers,
and secure session IDs. The transport guide also has a dedicated security
section for origin validation as DNS rebinding prevention and authentication.
The official MCP specification warns that HTTP transports need origin
validation to prevent DNS rebinding, should bind locally for local servers, and
should implement authentication. It also says that without those protections,
remote websites can interact with local MCP servers.
High Basis: CVSS v3 score 8.3, vector CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:C/C:H/I:H/A:H
0.6.0, fixed 1.6.59 (ecosystem)1.6.59
1.6.59pip install --upgrade praisonaiagents==1.6.59Fix commands are the export's own generic commands. Ojuju never runs them. Check them against your setup first.
Check your own computer: guard packages check
Source: PyPA Advisory Database · CC BY 4.0