Development

.github/release.yml

@@ -0,0 +1,50 @@
+# Configuration for GitHub's automatically generated release notes.
+#
+# When a release is drafted, every merged PR since the last tag is listed as
+# "* <PR title> by @<author> in #<PR>", so each contributor is credited with
+# their GitHub handle automatically, and a "New Contributors" section tags
+# first-time contributors. PRs are grouped into the categories below by label.
+#
+# Docs: https://docs.github.com/en/repositories/releasing-projects-on-github/automatically-generated-release-notes
+changelog:
+  exclude:
+    labels:
+      - ignore-for-release
+    authors:
+      - dependabot
+      - github-actions
+  categories:
+    - title: 🚨 Breaking Changes
+      labels:
+        - breaking-change
+    - title: ✨ Features
+      labels:
+        - enhancement
+    - title: 🐛 Bug Fixes
+      labels:
+        - bug
+    - title: ⚡ Reliability & Performance
+      labels:
+        - reliability
+        - performance
+    - title: 🧪 Tests
+      labels:
+        - testing
+    - title: 🏷️ Typing
+      labels:
+        - typing
+    - title: 🌊 Streaming & Observability
+      labels:
+        - streaming
+        - observability
+        - error-handling
+    - title: 📝 Documentation
+      labels:
+        - documentation
+    - title: 🔧 CI & Tooling
+      labels:
+        - ci
+        - chore
+    - title: 🔁 Other Changes
+      labels:
+        - "*"

CHANGELOG.md

@@ -7,6 +7,19 @@ This project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.htm
 
 ---
 
+## [Unreleased]
+
+### Fixed
+- Streaming completions now raise `FerroStreamError` on malformed SSE chunks
+  instead of silently dropping them, for both the sync and async clients. This
+  is the first code path that actually raises the previously-unused
+  `FerroStreamError`.
+
+### Added
+- Test coverage for `admin.dashboard()` and `admin.plugins.list()` (bare-array,
+  `{"data": [...]}`, and `{"plugins": [...]}` response shapes).
+- Streaming malformed-chunk regression tests for the sync and async clients.
+
 ## [0.2.0] — 2026-05-14
 
 ### Added

ferrolabsai/completions/async_resource.py

@@ -8,6 +8,7 @@
 
 import httpx
 
+from ..exceptions import FerroStreamError
 from ..types import ChatCompletion, ChatCompletionChunk
 
 
@@ -85,6 +86,9 @@ async def _stream(self, path: str, body: dict[str, Any]) -> AsyncIterator[ChatCo
                     if payload == "[DONE]":
                         return
                     try:
-                        yield ChatCompletionChunk.from_dict(json.loads(payload))
-                    except json.JSONDecodeError:
-                        continue
+                        chunk_data = json.loads(payload)
+                    except json.JSONDecodeError as e:
+                        raise FerroStreamError(
+                            f"Malformed SSE chunk in streaming response: {payload[:200]!r}"
+                        ) from e
+                    yield ChatCompletionChunk.from_dict(chunk_data)

ferrolabsai/completions/resource.py

@@ -6,6 +6,7 @@
 from collections.abc import Iterator
 from typing import Any, Literal, overload
 
+from ..exceptions import FerroStreamError
 from ..types import ChatCompletion, ChatCompletionChunk
 
 
@@ -137,6 +138,8 @@ def _stream(self, path: str, body: dict[str, Any]) -> Iterator[ChatCompletionChu
                     return
                 try:
                     chunk_data = json.loads(payload)
-                    yield ChatCompletionChunk.from_dict(chunk_data)
-                except json.JSONDecodeError:
-                    continue
+                except json.JSONDecodeError as e:
+                    raise FerroStreamError(
+                        f"Malformed SSE chunk in streaming response: {payload[:200]!r}"
+                    ) from e
+                yield ChatCompletionChunk.from_dict(chunk_data)

tests/test_sdk.py

@@ -20,6 +20,7 @@
     FerroNotFoundError,
     FerroRateLimitError,
     FerroServerError,
+    FerroStreamError,
 )
 
 BASE_URL = "http://localhost:8080"
@@ -497,6 +498,23 @@ def test_rollback(self, client, httpx_mock: HTTPXMock):
         assert result["rolled_back_to"] == 1
 
 
+class TestAdminDashboard:
+    def test_get_dashboard(self, client, httpx_mock: HTTPXMock):
+        httpx_mock.add_response(
+            method="GET",
+            url=f"{BASE_URL}/admin/dashboard",
+            json={
+                "providers": {"enabled": 3, "disabled": 1},
+                "keys": {"active": 5, "revoked": 2},
+                "requests": {"total": 128, "errors": 4},
+            },
+        )
+        dashboard = client.admin.dashboard()
+        assert dashboard["providers"]["enabled"] == 3
+        assert dashboard["keys"]["active"] == 5
+        assert dashboard["requests"]["total"] == 128
+
+
 class TestAdminLogs:
     def test_list_logs(self, client, httpx_mock: HTTPXMock):
         httpx_mock.add_response(
@@ -522,6 +540,41 @@ def test_logs_stats(self, client, httpx_mock: HTTPXMock):
         assert stats["total"] == 42
 
 
+class TestAdminPlugins:
+    def test_list_plugins_accepts_bare_array(self, client, httpx_mock: HTTPXMock):
+        plugins = [
+            {"name": "cache", "enabled": True},
+            {"name": "logger", "enabled": False},
+        ]
+        httpx_mock.add_response(
+            method="GET",
+            url=f"{BASE_URL}/admin/plugins",
+            json=plugins,
+        )
+
+        assert client.admin.plugins.list() == plugins
+
+    def test_list_plugins_accepts_data_wrapper(self, client, httpx_mock: HTTPXMock):
+        plugins = [{"name": "ratelimit", "enabled": True}]
+        httpx_mock.add_response(
+            method="GET",
+            url=f"{BASE_URL}/admin/plugins",
+            json={"data": plugins},
+        )
+
+        assert client.admin.plugins.list() == plugins
+
+    def test_list_plugins_accepts_plugins_wrapper(self, client, httpx_mock: HTTPXMock):
+        plugins = [{"name": "logger", "enabled": True}]
+        httpx_mock.add_response(
+            method="GET",
+            url=f"{BASE_URL}/admin/plugins",
+            json={"plugins": plugins},
+        )
+
+        assert client.admin.plugins.list() == plugins
+
+
 # ------------------------------------------------------------------
 # Error handling
 # ------------------------------------------------------------------
@@ -636,6 +689,40 @@ def test_sync_stream_500_raises_server_error(self, client, httpx_mock: HTTPXMock
                 )
             )
 
+    def test_sync_stream_malformed_chunk_raises_stream_error(self, client, httpx_mock: HTTPXMock):
+        sse_data = "data: {not valid json}\n\n"
+        httpx_mock.add_response(
+            method="POST",
+            url=f"{BASE_URL}/v1/chat/completions",
+            content=sse_data.encode(),
+        )
+        with pytest.raises(FerroStreamError, match="Malformed SSE chunk"):
+            list(
+                client.chat.completions.create(
+                    model="gpt-4o",
+                    messages=[{"role": "user", "content": "Hi"}],
+                    stream=True,
+                )
+            )
+
+    @pytest.mark.asyncio
+    async def test_async_stream_malformed_chunk_raises_stream_error(
+        self, async_client, httpx_mock: HTTPXMock
+    ):
+        sse_data = "data: {not valid json}\n\n"
+        httpx_mock.add_response(
+            method="POST",
+            url=f"{BASE_URL}/v1/chat/completions",
+            content=sse_data.encode(),
+        )
+        with pytest.raises(FerroStreamError, match="Malformed SSE chunk"):
+            async for _ in await async_client.chat.completions.create(
+                model="gpt-4o",
+                messages=[{"role": "user", "content": "Hi"}],
+                stream=True,
+            ):
+                pass
+
 
 # ------------------------------------------------------------------
 # P0-2: Async completions has parity params