Merge branch 'main' of ssh://ramanujan.glitch.university:2222/glitch-university/Agent0

agents/gunnar/settings.json

@@ -1,38 +1,39 @@
 {
-  "version": "v1.10",
-  "api_keys": {},
-  "auth_login": "",
-  "auth_password": "",
-  "root_password": "",
-  "agent_profile": "agent0",
-  "agent_knowledge_subdir": "custom",
-  "workdir_path": "/a0/usr/workdir",
-  "workdir_show": true,
-  "workdir_max_depth": 5,
-  "workdir_max_files": 20,
-  "workdir_max_folders": 20,
-  "workdir_max_lines": 250,
-  "workdir_gitignore": "# Python environments & cache\nvenv/**\n**/__pycache__/**\n\n# Node.js dependencies\n**/node_modules/**\n**/.npm/**\n\n# Version control metadata\n**/.git/**",
-  "rfc_auto_docker": true,
-  "rfc_url": "localhost",
-  "rfc_password": "",
-  "rfc_port_http": 55080,
-  "websocket_server_restart_enabled": true,
-  "uvicorn_access_logs_enabled": false,
-  "stt_model_size": "base",
-  "stt_language": "en",
-  "stt_silence_threshold": 0.3,
-  "stt_silence_duration": 1000,
-  "stt_waiting_timeout": 2000,
-  "tts_kokoro": true,
-  "mcp_servers": "{\n    \"mcpServers\": {}\n}",
-  "mcp_client_init_timeout": 10,
-  "mcp_client_tool_timeout": 120,
-  "mcp_server_enabled": true,
-  "a2a_server_enabled": false,
-  "variables": "",
-  "secrets": "",
-  "litellm_global_kwargs": {},
-  "update_check_enabled": true,
-  "chat_inherit_project": true
+    "version": "v1.10",
+    "api_keys": {"gnommoweb": "RVWUl_-I1W3EjDpM"},
+    "auth_login": "",
+    "auth_password": "",
+    "root_password": "",
+    "agent_profile": "agent0",
+    "agent_knowledge_subdir": "custom",
+    "workdir_path": "/a0/usr/workdir",
+    "workdir_show": true,
+    "workdir_max_depth": 5,
+    "workdir_max_files": 20,
+    "workdir_max_folders": 20,
+    "workdir_max_lines": 250,
+    "workdir_gitignore": "# Python environments & cache\nvenv/**\n**/__pycache__/**\n\n# Node.js dependencies\n**/node_modules/**\n**/.npm/**\n\n# Version control metadata\n**/.git/**",
+    "rfc_auto_docker": true,
+    "rfc_url": "localhost",
+    "rfc_password": "",
+    "rfc_port_http": 55080,
+    "websocket_server_restart_enabled": true,
+    "uvicorn_access_logs_enabled": false,
+    "stt_model_size": "base",
+    "stt_language": "en",
+    "stt_silence_threshold": 0.3,
+    "stt_silence_duration": 1000,
+    "stt_waiting_timeout": 2000,
+    "tts_kokoro": true,
+    "mcp_servers": "{\n    \"mcpServers\": {}\n}",
+    "mcp_client_init_timeout": 10,
+    "mcp_client_tool_timeout": 120,
+    "mcp_server_enabled": false,
+    "mcp_server_token": "",
+    "a2a_server_enabled": false,
+    "variables": "",
+    "secrets": "",
+    "litellm_global_kwargs": {},
+    "update_check_enabled": true,
+    "chat_inherit_project": true
 }

plugins/festinger/festinger/main.py

@@ -2162,52 +2162,52 @@ async def update_config(request: Request) -> dict:
 
 @app.get("/test-chat/models")
 async def test_chat_models(request: Request) -> dict:
-    """Return the list of models currently available in upstream_openai (Ollama)."""
+    """Return the list of lm-studio/openai models configured in the DB."""
+    pool = request.app.state.pool
     cfg = request.app.state.yaml_config
     upstream = cfg.get("upstream_openai", "")
-    if not upstream:
-        return {"ok": False, "error": "upstream_openai not configured", "models": []}
-    url = f"{upstream.rstrip('/')}/v1/models"
-    try:
-        async with httpx.AsyncClient(timeout=8.0) as client:
-            r = await client.get(url)
-        if not r.is_success:
-            return {"ok": False, "error": f"HTTP {r.status_code} from {url}", "models": []}
-        ids = sorted(m.get("id", "") for m in r.json().get("data", []))
-        return {"ok": True, "models": ids, "upstream": upstream}
-    except httpx.ConnectError:
-        return {"ok": False, "error": f"Connection refused at {url} — is Ollama running?", "models": []}
-    except Exception as exc:
-        return {"ok": False, "error": str(exc), "models": []}
+    async with pool.acquire() as conn:
+        rows = await conn.fetch(
+            "SELECT model_name, base_url FROM models WHERE provider IN ('lm-studio', 'openai') ORDER BY id"
+        )
+    if rows:
+        models = [r["model_name"] for r in rows]
+        return {"ok": True, "models": models, "upstream": upstream}
+    return {"ok": False, "error": "No lm-studio or openai models configured — add one via Model Manager.", "models": []}
 
 
 @app.post("/test-chat")
 async def test_chat(request: Request) -> dict:
     cfg = request.app.state.yaml_config
+    pool = request.app.state.pool
     data = await request.json()
-    message  = (data.get("message") or "Say hello in one sentence.").strip()
-    model    = (data.get("model")   or "").strip()
+    message = (data.get("message") or "Say hello in one sentence.").strip()
+    model   = (data.get("model")   or "").strip()
     upstream = cfg.get("upstream_openai", "")
 
-    if not upstream:
-        return {"ok": False, "error": "upstream_openai not configured"}
-
-    # Always fetch the available model list so the UI can show it
-    available: list[str] = []
-    try:
-        async with httpx.AsyncClient(timeout=8.0) as client:
-            mr = await client.get(f"{upstream.rstrip('/')}/v1/models")
-        if mr.is_success:
-            available = sorted(m.get("id", "") for m in mr.json().get("data", []))
-    except Exception:
-        pass
+    # Load configured models from DB so the UI can refresh the dropdown
+    async with pool.acquire() as conn:
+        db_rows = await conn.fetch(
+            "SELECT model_name, base_url FROM models WHERE provider IN ('lm-studio', 'openai') ORDER BY id"
+        )
+    available = [r["model_name"] for r in db_rows]
+    db_model_map = {r["model_name"]: r["base_url"] or "" for r in db_rows}
 
     if not model:
         if not available:
-            return {"ok": False, "error": f"No model specified and none found at {upstream}/v1/models", "available": []}
+            return {"ok": False, "error": "No lm-studio/openai models configured. Add one via Model Manager first.", "available": []}
         model = available[0]
 
-    url = f"{upstream.rstrip('/')}/v1/chat/completions"
+    # Resolve upstream: prefer the model's own base_url from DB, fall back to upstream_openai config
+    raw_base = db_model_map.get(model, "") or upstream
+    if not raw_base:
+        return {"ok": False, "error": "upstream_openai not configured and model has no base_url", "available": available}
+    # Strip /v1 suffix — we append /v1/chat/completions ourselves
+    effective_upstream = raw_base.rstrip("/")
+    if effective_upstream.endswith("/v1"):
+        effective_upstream = effective_upstream[:-3]
+
+    url = f"{effective_upstream}/v1/chat/completions"
     body = {
         "model": model,
         "stream": False,
@@ -2222,11 +2222,12 @@ async def test_chat(request: Request) -> dict:
         if not r.is_success:
             return {"ok": False, "error": f"HTTP {r.status_code}: {r.text[:400]}", "url": url, "model": model, "ms": ms, "available": available}
         resp = r.json()
-        reply = resp.get("choices", [{}])[0].get("message", {}).get("content", "(empty)")
+        choices = resp.get("choices") or []
+        reply = choices[0].get("message", {}).get("content", "(empty)") if choices else "(empty response)"
         return {"ok": True, "reply": reply, "model": model, "url": url, "ms": ms, "available": available}
     except httpx.ConnectError as exc:
         ms = int((time.perf_counter() - t0) * 1000)
-        return {"ok": False, "error": f"Connection refused — is Ollama running at {upstream}? ({exc})", "url": url, "ms": ms, "available": available}
+        return {"ok": False, "error": f"Connection refused at {effective_upstream} ({exc})", "url": url, "ms": ms, "available": available}
     except httpx.TimeoutException:
         ms = int((time.perf_counter() - t0) * 1000)
         return {"ok": False, "error": f"Request timed out after {ms}ms", "url": url, "ms": ms, "available": available}
@@ -3705,7 +3706,7 @@ ADMIN_HTML = """<!DOCTYPE html>
 
   <h2>Test Chat</h2>
   <p style="font-size:0.83em;color:#666;margin-bottom:0.8em">
-    Send a message directly to <code>upstream_openai</code> (Ollama) to verify the connection end-to-end.
+    Send a test message via a configured model (lm-studio / openai) to verify the connection end-to-end.
   </p>
   <div id="tc-models-row" style="font-size:0.82em;color:#555;margin-bottom:0.6em">
     <span id="tc-models-label">Loading available models…</span>