🛡️ Sentinel: [CRITICAL] Fix missing SSRF protection in public proof bundle example

.jules/sentinel.md

@@ -0,0 +1,4 @@
+## 2023-10-24 - [CRITICAL] Prevent SSRF in urllib.request.urlopen
+**Vulnerability:** The functions `_fetch_live_manifest` and `_fetch_live_runtime_proof` in `examples/generate_public_proof_bundle.py` use `urllib.request.urlopen` to fetch data from URLs provided via arguments. There was no validation to ensure the URL scheme was `http` or `https`. This could allow an attacker to pass a `file://` or `ftp://` scheme, potentially reading arbitrary local files or making unexpected requests.
+**Learning:** Python's `urllib.request.urlopen` handles various protocols like `file://` by default. When using URLs from unverified sources, we must explicitly whitelist acceptable schemes.
+**Prevention:** Always validate that URLs start with `http://` or `https://` (case-insensitive) before passing them to `urllib.request.urlopen` or similar functions.

examples/generate_public_proof_bundle.py

@@ -94,6 +94,17 @@ def _public_manifest_receipt(payload):
 
 
 def _fetch_live_manifest(url):
+    if not url or not (url.lower().startswith('http://') or url.lower().startswith('https://')):
+        return {
+            'included': False,
+            'attempted': True,
+            'route': url,
+            'http_status': 0,
+            'fetched_at': datetime.now(timezone.utc).replace(microsecond=0).isoformat(),
+            'body_sha256': '',
+            'error': f"Invalid URL scheme in {url!r}",
+            'manifest': {},
+        }
     try:
         with urllib.request.urlopen(url, timeout=10) as response:
             raw = response.read()
@@ -124,6 +135,17 @@ def _derive_sibling_url(url, old_suffix, new_suffix):
 
 
 def _fetch_live_runtime_proof(url):
+    if not url or not (url.lower().startswith('http://') or url.lower().startswith('https://')):
+        return {
+            'included': False,
+            'attempted': True,
+            'route': url,
+            'http_status': 0,
+            'fetched_at': datetime.now(timezone.utc).replace(microsecond=0).isoformat(),
+            'body_sha256': '',
+            'error': f"Invalid URL scheme in {url!r}",
+            'receipt': {},
+        }
     try:
         with urllib.request.urlopen(url, timeout=15) as response:
             raw = response.read()