GHSA-wv94-5qcp-6m36: Medium severity pip/PraisonAI vulnerability
Summary
The PraisonAI MCP HTTP-stream server creates a new in-memory session on every initialize request and never removes it. The cleanup routine that would expire sessions (cleanupsessions) is defined but never called anywhere in the codebase, and the configured session TTL is never enforced. There is no cap on the number of sessions. Because initialize requires no authentication and the server keeps every session dictionary forever, an attacker who can reach the endpoint (directly when the server is bound to a routable address, or from a victim's browser via the separate Origin-validation bypass) can drive memory usage up without bound until the process is killed by the out-of-memory killer. The same unbounded-growth pattern also applies to the cancelled-requests set populated by notifications/cancelled.
Details
In transports/httpstream.py, each initialize creates and stores a session with no limit:
python if body.get("method") == "initialize": newsessionid = str(uuid.uuid4()) self.sessions[newsessionid] = { "createdat": time.time(), "lastactivity": time.time(), }
A cleanup method exists:
python def cleanupsessions(self) -> None: now = time.time() expired = [sid for sid, data in self.sessions.items() if now - data["lastactivity"] > self.sessionttl] for sid in expired: del self.sessions[sid]
but grep across the package shows it has no call sites: it is never invoked on a timer, on request handling, or from any background task. self.sessionttl (default 3600) is stored and otherwise unused. There is no maximum-session check anywhere on the write path. As a result self.sessions grows monotonically for the lifetime of the process.
initialize is unauthenticated: in mcppost the API-key check is skipped when no key is configured (the default), and initialize does not require a prior session. The Origin check is the only gate, and a request with no Origin header is allowed; additionally the Origin allowlist is bypassable (see the companion report on the startswith Origin-validation bypass), so the endpoint is reachable from a malicious web page as well as directly.
The server-side cancellation set in server.py has the same defect:
python if method == "notifications/cancelled": requestid = params.get("requestId") if requestid: self.cancelledrequests.add(str(requestid)) # never cleared
self.cancelledrequests is an unbounded set that is added to but never pruned.
PoC
scripts/pocmcpsessiondos.sh. Start the server (default config, no API key):
praisonai mcp serve --transport http-stream --host 127.0.0.1 --port 8080
Send repeated initialize requests and watch the active session count grow:
bash for i in $(seq 1 200); do curl -s -o /dev/null -X POST http://127.0.0.1:8080/mcp \ -H 'Content-Type: application/json' -H 'Origin: http://localhost' \ -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-06-18","capabilities":{},"clientInfo":{"name":"x","version":"1"}}}' done curl -s http://127.0.0.1:8080/health
Observed on 4.6.52 after 200 requests:
{"status":"healthy","server":"praisonai","version":"1.0.0","protocolversion":"2025-11-25","activesessions":200}
The count rises by one per request and never decreases; there is no TTL expiry and no cap. Sustained requests grow the process resident set without bound. Each session also retains any SSE event history keyed by session id, amplifying the per-session footprint.
Impact
An unauthenticated client can exhaust the memory of the host running the MCP server, leading to denial of service (the process is terminated by the OOM killer, taking down the agent endpoint). When the server is bound to a routable interface (for example --host 0.0.0.0, common in containers), this is a direct remote unauthenticated DoS. With the default localhost bind, it is reachable from any web page the operator visits, because initialize is unauthenticated and the Origin gate is bypassable. The defect is a missing cleanup wiring plus the absence of any session cap, so it manifests even under benign long-running use.
Remediation
Enforce the session TTL and cap the number of concurrent sessions: call cleanupsessions periodically (a background asyncio task, or opportunistically on each request) and reject new sessions with a 429/503 once a configurable maximum is reached. Bound cancelledrequests similarly (for example an LRU or a periodic prune keyed by age), since it is also never cleared. Require authentication by default on the HTTP-stream transport so that anonymous clients cannot create sessions at all.
Affected Software
Remediation
Recommended actions to resolve this vulnerability, in priority order.
- Upgrade
Upgrade
pip/PraisonAIto a version that resolves this vulnerability.Fixed in 4.6.58 - Configuration
Wire session expiration so _cleanup_sessions is invoked (e.g., a background asyncio task and/or opportunistically on request handling). Ensure entries in _sessions are removed when now - data["last_activity"] > self.session_ttl (default 3600).
PraisonAI MCP HTTP-stream transport (transports/http_stream.py) session_ttl = Enforce configured TTL (default 3600) and cap sessions/cleanup by calling _cleanup_sessions periodically or on each request - Configuration
Implement a session cap: before storing a new session created by initialize, check the current active session count against a configurable maximum; if the maximum is reached, reject the request with HTTP 429/503 instead of adding to self._sessions.
PraisonAI MCP HTTP-stream transport (transports/http_stream.py) max concurrent sessions = Add configurable maximum and reject when reached with 429/503 - Configuration
Require authentication by default on the HTTP-stream transport so anonymous clients cannot create sessions. This prevents unauthenticated initialize requests from exhausting host memory.
PraisonAI MCP HTTP-stream transport (transports/http_stream.py) authentication requirement for /mcp (HTTP-stream) = Require authentication by default - Configuration
Implement pruning/capping for self._cancelled_requests (populated by notifications/cancelled). Ensure request IDs are eventually removed instead of growing unbounded (e.g., TTL/age-based prune similar to session cleanup, or a max-size bounded LRU/periodic prune).
PraisonAI MCP HTTP-stream transport (server.py / cancellation handling) cancelled requests pruning = Add cap/TTL prune for _cancelled_requests
Event History
Frequently Asked Questions
Which deployments are exposed to unauthenticated resource exhaustion?
Any HTTP-stream server reachable by an attacker is exposed because initialize requires no authentication. This includes servers bound to routable addresses and servers reachable through a victim's browser using the separate Origin-validation bypass.
What does an attacker need to do to exhaust the service?
The attacker can repeatedly send initialize requests, each of which creates a new in-memory session. They can also cause unbounded growth in the cancelled-requests set through notifications/cancelled.
What is the operational impact of successful exploitation?
Memory use can grow without bound because sessions are retained indefinitely and there is no session-count cap. The process may eventually be terminated by the out-of-memory killer, causing a denial of service.
Is the configured session TTL an effective protection?
No. Although a session cleanup routine exists, it is never called, so the configured session TTL is not enforced.
Is a release associated with a fix available?
The advisory references PraisonAI release v4.6.58 and commit 2f9677abb2ea68eab864ee8b6a828fd0141612e1.