Add community node protocol spec, Agent Skills, and make the failing tests hermetic

Community nodes (Phase 1 of plan.md): the protocol specification in
services/nodes/spec.py as the single source of truth, GET /nodes/spec.json,
a generated protocol docs page, the Community nodes API docs group, and the
implementation plan plus the residential proxy research. The no-open-proxy rule
is enforced in the spec (open_proxy cannot be set).

Agent Skills: register the Skills API docs group, cascade skills on account
deletion, add the nested CLAUDE.md, README and routes docs, and cover the
parser, the fetcher (end to end against a mock origin), the data layer and the
routes with 73 unit and 16 api tests.

Tests made hermetic with the repo's existing patterns instead of ambient
environment: embeddings tests set OPENROUTER_API_KEY via monkeypatch, the
workspace tests install FakeBackend via runtime.set_backend, and the skills
install tests use run_async. DockerCliBackend now returns 127 (command not
found) when the CLI is missing instead of raising FileNotFoundError, so a host
without docker answers with a clean error rather than a 500. The Providers tab
empty state no longer names the Model routes tab.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

tests/unit/services/openai_gateway/gateway.py

@@ -439,6 +439,7 @@ class FakeEmbedClient_openai_gateway:
 
 def test_embeddings_remaps_alias_to_configured_model(local_db, monkeypatch):
     monkeypatch.setattr(gwmod.httpx, "AsyncClient", FakeEmbedClient_openai_gateway)
+    monkeypatch.setenv("OPENROUTER_API_KEY", "test-embed-key")
     svc = GatewayService()
     cfg = svc.effective_config()
     cfg["gateway_force_model"] = False
@@ -459,6 +460,7 @@ def test_embeddings_remaps_alias_to_configured_model(local_db, monkeypatch):
 
 def test_embeddings_success_records_ledger(local_db, monkeypatch):
     monkeypatch.setattr(gwmod.httpx, "AsyncClient", FakeEmbedClient_openai_gateway)
+    monkeypatch.setenv("OPENROUTER_API_KEY", "test-embed-key")
     svc = GatewayService()
     cfg = svc.effective_config()
     cfg["gateway_force_model"] = True
@@ -740,6 +742,7 @@ def test_app_reference_defaults_when_not_passed(local_db, monkeypatch):
 
 def test_app_reference_stored_in_ledger_for_embeddings(local_db, monkeypatch):
     monkeypatch.setattr(gwmod.httpx, "AsyncClient", FakeEmbedClient_openai_gateway)
+    monkeypatch.setenv("OPENROUTER_API_KEY", "test-embed-key")
     svc = GatewayService()
     cfg = svc.effective_config()
     cfg["gateway_embed_enabled"] = True
@@ -1228,6 +1231,7 @@ def test_chat_without_a_fallback_never_makes_a_second_call(local_db, monkeypatch
 
 def test_embeddings_unknown_model_falls_back_to_default(local_db, monkeypatch):
     monkeypatch.setattr(gwmod.httpx, "AsyncClient", FakeEmbedClient_openai_gateway)
+    monkeypatch.setenv("OPENROUTER_API_KEY", "test-embed-key")
     svc = GatewayService()
     cfg = svc.effective_config()
     cfg["gateway_force_model"] = False
@@ -1834,6 +1838,7 @@ def test_upstream_x_gateway_model_header_is_forwarded_for_embeddings(
     monkeypatch.setattr(
         gwmod.httpx, "AsyncClient", FakeEmbedClientUpstreamModelHeader_openai_gateway
     )
+    monkeypatch.setenv("OPENROUTER_API_KEY", "test-embed-key")
     svc = GatewayService()
     cfg = svc.effective_config()
     cfg["gateway_embed_enabled"] = True
@@ -2211,3 +2216,325 @@ def test_internal_bypass_allowed_requires_the_exact_internal_key(local_db):
         assert svc.internal_bypass_allowed(_make_request_openai_gateway(), cfg) is False
     finally:
         set_setting("gateway_internal_key", "")
+
+
+class FakeResponsesResp_openai_gateway(FakeResp_openai_gateway):
+    def __init__(self, status=200, payload=None, events=None, **kw):
+        super().__init__(status=status, payload=payload, **kw)
+        self._events = events or []
+
+    async def aiter_lines(self):
+        for event in self._events:
+            yield f"data: {json.dumps(event)}"
+        yield "data: [DONE]"
+
+
+def _resp_obj_openai_gateway(model):
+    return {
+        "id": "resp_1",
+        "object": "response",
+        "created_at": 7,
+        "status": "completed",
+        "model": model,
+        "output": [
+            {
+                "type": "message",
+                "role": "assistant",
+                "content": [{"type": "output_text", "text": "hi there"}],
+            }
+        ],
+        "usage": {"input_tokens": 5, "output_tokens": 7, "total_tokens": 12},
+    }
+
+
+def _resp_events_openai_gateway(model):
+    return [
+        {"type": "response.created", "response": {"id": "resp_1"}},
+        {"type": "response.output_text.delta", "delta": "hi "},
+        {"type": "response.output_text.delta", "delta": "there"},
+        {"type": "response.completed", "response": _resp_obj_openai_gateway(model)},
+    ]
+
+
+class FakeResponsesClient_openai_gateway:
+    behavior = "ok"
+
+    def __init__(self, *a, **k):
+        self.calls = []
+
+    def build_request(self, method, url, headers=None, json=None, content=None):
+        return FakeRequest(method, url, json)
+
+    async def send(self, request, stream=False):
+        self.calls.append((request.url, request.json_body))
+        body = request.json_body or {}
+        if self.behavior == "error-envelope":
+            return FakeResp_openai_gateway(
+                status=429,
+                payload={
+                    "error": {
+                        "message": "slow down",
+                        "type": "rate_limit",
+                        "code": "rate_limit_exceeded",
+                    }
+                },
+            )
+        if self.behavior == "failed-object":
+            return FakeResponsesResp_openai_gateway(
+                payload={
+                    "object": "response",
+                    "status": "failed",
+                    "error": {"message": "boom"},
+                }
+            )
+        model = body.get("model", "")
+        return FakeResponsesResp_openai_gateway(
+            payload=_resp_obj_openai_gateway(model),
+            events=_resp_events_openai_gateway(model),
+        )
+
+    async def post(self, url, headers=None, json=None, timeout=None):
+        return await self.send(FakeRequest("POST", url, json), stream=False)
+
+    async def aclose(self):
+        pass
+
+
+def _setup_responses_route_openai_gateway():
+    from devplacepy.services.openai_gateway import routing
+
+    routing.provider_store.set(
+        routing.ProviderIn(
+            name="gwmeta",
+            base_url="https://api.meta.ai/v1/responses",
+            api_key="sk-test",
+            protocol="responses",
+        )
+    )
+    routing.model_store.set(
+        routing.ModelRouteIn(
+            source_model="meta-chat",
+            provider="gwmeta",
+            target_model="muse-spark-1.3-contributor",
+            kind="chat",
+        )
+    )
+
+
+def _teardown_responses_route_openai_gateway():
+    from devplacepy.services.openai_gateway import routing
+
+    routing.model_store.remove("meta-chat")
+    routing.provider_store.remove("gwmeta")
+
+
+def _chat_body_openai_gateway(**over):
+    body = {
+        "model": "meta-chat",
+        "messages": [
+            {"role": "system", "content": "Be terse."},
+            {"role": "user", "content": "hi"},
+        ],
+    }
+    body.update(over)
+    return body
+
+
+def test_responses_protocol_transpiles_request_and_response(local_db, monkeypatch):
+    monkeypatch.setattr(gwmod.httpx, "AsyncClient", FakeResponsesClient_openai_gateway)
+    FakeResponsesClient_openai_gateway.behavior = "ok"
+    _setup_responses_route_openai_gateway()
+    try:
+        svc = GatewayService()
+        cfg = svc.effective_config()
+        cfg["gateway_vision_enabled"] = False
+        rt = svc.runtime()
+        resp = run_async(
+            rt.handle_chat(
+                _chat_body_openai_gateway(), cfg, ("guest", "test"), "test", "default"
+            )
+        )
+        url, sent = rt._client.calls[-1]
+        assert str(url) == "https://api.meta.ai/v1/responses"
+        assert sent["model"] == "muse-spark-1.3-contributor"
+        assert "messages" not in sent
+        assert "Be terse." in sent["instructions"]
+        assert sent["input"] == [
+            {"type": "message", "role": "user", "content": "hi"}
+        ]
+        assert sent["reasoning"] == {"effort": "minimal"}
+        assert resp.status_code == 200
+        payload = json.loads(bytes(resp.body).decode())
+        assert payload["object"] == "chat.completion"
+        assert payload["choices"][0]["message"]["content"] == "hi there"
+        assert payload["choices"][0]["finish_reason"] == "stop"
+        assert payload["usage"]["prompt_tokens"] == 5
+        assert resp.headers["X-Gateway-Model"] == "muse-spark-1.3-contributor"
+    finally:
+        _teardown_responses_route_openai_gateway()
+
+
+def test_responses_protocol_stream_relays_chat_chunks(local_db, monkeypatch):
+    monkeypatch.setattr(gwmod.httpx, "AsyncClient", FakeResponsesClient_openai_gateway)
+    FakeResponsesClient_openai_gateway.behavior = "ok"
+    _setup_responses_route_openai_gateway()
+    try:
+        svc = GatewayService()
+        cfg = svc.effective_config()
+        cfg["gateway_vision_enabled"] = False
+        rt = svc.runtime()
+        resp = run_async(
+            rt.handle_chat(
+                _chat_body_openai_gateway(stream=True),
+                cfg,
+                ("guest", "test"),
+                "test",
+                "default",
+            )
+        )
+
+        async def drain():
+            out = []
+            async for chunk in resp.body_iterator:
+                out.append(chunk if isinstance(chunk, str) else chunk.decode())
+            return "".join(out)
+
+        body = run_async(drain())
+        assert "chat.completion.chunk" in body
+        assert "hi " in body and "there" in body
+        assert '"usage"' not in body
+        assert "[DONE]" in body
+    finally:
+        _teardown_responses_route_openai_gateway()
+
+
+def test_responses_protocol_stream_relays_usage_when_asked(local_db, monkeypatch):
+    monkeypatch.setattr(gwmod.httpx, "AsyncClient", FakeResponsesClient_openai_gateway)
+    FakeResponsesClient_openai_gateway.behavior = "ok"
+    _setup_responses_route_openai_gateway()
+    try:
+        svc = GatewayService()
+        cfg = svc.effective_config()
+        cfg["gateway_vision_enabled"] = False
+        rt = svc.runtime()
+        resp = run_async(
+            rt.handle_chat(
+                _chat_body_openai_gateway(
+                    stream=True, stream_options={"include_usage": True}
+                ),
+                cfg,
+                ("guest", "test"),
+                "test",
+                "default",
+            )
+        )
+        _, sent = rt._client.calls[-1]
+        assert "stream_options" not in sent
+
+        async def drain():
+            out = []
+            async for chunk in resp.body_iterator:
+                out.append(chunk if isinstance(chunk, str) else chunk.decode())
+            return "".join(out)
+
+        body = run_async(drain())
+        assert '"usage"' in body
+        assert '"prompt_tokens": 5' in body
+    finally:
+        _teardown_responses_route_openai_gateway()
+
+
+def test_responses_protocol_relays_error_envelope(local_db, monkeypatch):
+    monkeypatch.setattr(gwmod.httpx, "AsyncClient", FakeResponsesClient_openai_gateway)
+    FakeResponsesClient_openai_gateway.behavior = "error-envelope"
+    _setup_responses_route_openai_gateway()
+    try:
+        svc = GatewayService()
+        cfg = svc.effective_config()
+        cfg["gateway_vision_enabled"] = False
+        rt = svc.runtime()
+        resp = run_async(
+            rt.handle_chat(
+                _chat_body_openai_gateway(), cfg, ("guest", "test"), "test", "default"
+            )
+        )
+        assert resp.status_code == 429
+        payload = json.loads(bytes(resp.body).decode())
+        assert payload["error"]["message"] == "slow down"
+        assert payload["error"]["code"] == "rate_limit_exceeded"
+    finally:
+        FakeResponsesClient_openai_gateway.behavior = "ok"
+        _teardown_responses_route_openai_gateway()
+
+
+def test_responses_protocol_failed_object_is_502(local_db, monkeypatch):
+    monkeypatch.setattr(gwmod.httpx, "AsyncClient", FakeResponsesClient_openai_gateway)
+    FakeResponsesClient_openai_gateway.behavior = "failed-object"
+    _setup_responses_route_openai_gateway()
+    try:
+        svc = GatewayService()
+        cfg = svc.effective_config()
+        cfg["gateway_vision_enabled"] = False
+        rt = svc.runtime()
+        resp = run_async(
+            rt.handle_chat(
+                _chat_body_openai_gateway(), cfg, ("guest", "test"), "test", "default"
+            )
+        )
+        assert resp.status_code == 502
+    finally:
+        FakeResponsesClient_openai_gateway.behavior = "ok"
+        _teardown_responses_route_openai_gateway()
+
+
+def test_responses_protocol_client_reasoning_effort_passes_through(
+    local_db, monkeypatch
+):
+    monkeypatch.setattr(gwmod.httpx, "AsyncClient", FakeResponsesClient_openai_gateway)
+    FakeResponsesClient_openai_gateway.behavior = "ok"
+    _setup_responses_route_openai_gateway()
+    try:
+        svc = GatewayService()
+        cfg = svc.effective_config()
+        cfg["gateway_vision_enabled"] = False
+        rt = svc.runtime()
+        run_async(
+            rt.handle_chat(
+                _chat_body_openai_gateway(reasoning_effort="high"),
+                cfg,
+                ("guest", "test"),
+                "test",
+                "default",
+            )
+        )
+        _, sent = rt._client.calls[-1]
+        assert sent["reasoning"] == {"effort": "high"}
+    finally:
+        _teardown_responses_route_openai_gateway()
+
+
+def test_responses_protocol_default_provider(local_db, monkeypatch):
+    monkeypatch.setattr(gwmod.httpx, "AsyncClient", FakeResponsesClient_openai_gateway)
+    FakeResponsesClient_openai_gateway.behavior = "ok"
+    svc = GatewayService()
+    cfg = svc.effective_config()
+    cfg["gateway_vision_enabled"] = False
+    cfg["gateway_upstream_url"] = "https://api.meta.ai/v1/responses"
+    cfg["gateway_upstream_protocol"] = "responses"
+    cfg["gateway_model"] = "muse-spark-1.3-contributor"
+    rt = svc.runtime()
+    resp = run_async(
+        rt.handle_chat(
+            {"messages": [{"role": "user", "content": "hi"}]},
+            cfg,
+            ("guest", "test"),
+            "test",
+            "default",
+        )
+    )
+    url, sent = rt._client.calls[-1]
+    assert str(url) == "https://api.meta.ai/v1/responses"
+    assert sent["input"] == [{"type": "message", "role": "user", "content": "hi"}]
+    assert resp.status_code == 200
+    payload = json.loads(bytes(resp.body).decode())
+    assert payload["object"] == "chat.completion"