fix: preserve adapter failure diagnostics

.github/workflows/codeql.yml

@@ -4,10 +4,14 @@ on:
     branches: [ "main", "master" ]
   pull_request:
     branches: [ "main", "master" ]
+permissions:
+  contents: read
+
 jobs:
   analyze:
     name: Analyze
     runs-on: ubuntu-latest
+    timeout-minutes: 15
     permissions:
       actions: read
       contents: read

.github/workflows/pages.yml

@@ -14,6 +14,7 @@ concurrency:
 jobs:
   build:
     runs-on: ubuntu-latest
+    timeout-minutes: 10
     steps:
       - uses: actions/checkout@v4
       - uses: actions/setup-python@v5
@@ -29,6 +30,7 @@ jobs:
       name: github-pages
       url: ${{ steps.deployment.outputs.page_url }}
     runs-on: ubuntu-latest
+    timeout-minutes: 10
     needs: build
     steps:
       - name: Deploy to GitHub Pages

.github/workflows/pr-lint.yml

@@ -6,10 +6,15 @@ on:
       - edited
       - synchronize
 
+permissions:
+  contents: read
+  pull-requests: read
+
 jobs:
   main:
     name: Validate PR title
     runs-on: ubuntu-latest
+    timeout-minutes: 10
     steps:
       - uses: amannn/action-semantic-pull-request@v5
         env:

.github/workflows/stale.yml

@@ -3,9 +3,15 @@ on:
   schedule:
   - cron: '0 0 * * *'
 
+permissions:
+  contents: read
+  issues: write
+  pull-requests: write
+
 jobs:
   stale:
     runs-on: ubuntu-latest
+    timeout-minutes: 10
     steps:
     - uses: actions/stale@v8
       with:

src/cas_evals/reference_product.py

@@ -47,9 +47,9 @@ def post(envelope: dict[str, Any]) -> dict[str, Any]:
             with urlopen(request, timeout=timeout_seconds) as response:
                 payload = response.read(MAX_RESPONSE_BYTES + 1)
         except HTTPError as error:
-            raise ReferenceProductError(f"reference product returned HTTP {error.code}") from None
-        except (URLError, TimeoutError, OSError):
-            raise ReferenceProductError("reference product is unavailable") from None
+            raise ReferenceProductError(f"reference product returned HTTP {error.code}") from error
+        except (URLError, TimeoutError, OSError) as error:
+            raise ReferenceProductError("reference product is unavailable") from error
         if len(payload) > MAX_RESPONSE_BYTES:
             raise ReferenceProductError("reference product response exceeds the size limit")
         try:

tests/test_reference_product.py

@@ -6,8 +6,14 @@
 import unittest
 from http.server import BaseHTTPRequestHandler, ThreadingHTTPServer
 from pathlib import Path
+from unittest.mock import patch
+from urllib.error import URLError
 
-from cas_evals.reference_product import ReferenceProductError, evaluate_reference_suite
+from cas_evals.reference_product import (
+    ReferenceProductError,
+    _http_transport,
+    evaluate_reference_suite,
+)
 
 ROOT = Path(__file__).parents[1]
 GOLDEN = ROOT / "benchmarks/reference-product/v0.1/golden.json"
@@ -105,6 +111,18 @@ def test_invalid_endpoint_and_timeout_fail_closed(self):
         with self.assertRaises(ReferenceProductError):
             evaluate_reference_suite(GOLDEN, timeout_seconds=0)
 
+    def test_network_failure_preserves_root_cause(self):
+        failure = URLError("network unavailable")
+        transport = _http_transport("http://127.0.0.1:8080/api/v1/workflows", 1.0)
+
+        with patch("cas_evals.reference_product.urlopen", side_effect=failure):
+            with self.assertRaisesRegex(
+                ReferenceProductError, "reference product is unavailable"
+            ) as caught:
+                transport({"kind": "PromptEnvelope"})
+
+        self.assertIs(caught.exception.__cause__, failure)
+
     def test_http_endpoint_is_executable(self):
         server = ThreadingHTTPServer(("127.0.0.1", 0), ReferenceHandler)
         thread = threading.Thread(target=server.serve_forever, daemon=True)