LLM Team UI v1.0 — full-stack local AI orchestration platform

Features:
- 20 team modes (brainstorm, debate, consensus, red team, etc.)
- 3 autonomous pipelines (research, model eval, knowledge extraction)
- AutoResearch Lab with ratchet engine (Karpathy-inspired)
- Multi-provider support (Ollama, OpenRouter, OpenAI, Anthropic)
- Admin panel (providers, models, timeouts, OpenRouter browser)
- History panel with copy/iterate/re-pipe workflow
- Context budget system (smart truncation, safe_query, overflow recovery)
- PostgreSQL persistence (team_runs, pipeline_runs, lab_experiments, lab_trials)
- Pure Python + embedded HTML/CSS/JS, no external JS dependencies
- Inline SVG score charts in Lab monitor
- SSE streaming for real-time output
- Systemd service with auto-restart

Stack: Flask + Ollama + PostgreSQL + Bun-compatible
Hardware: RTX A4000 (16GB) + 128GB RAM

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

llm_team_config.json

@@ -0,0 +1,33 @@
+{
+  "providers": {
+    "ollama": {
+      "enabled": true,
+      "base_url": "http://localhost:11434",
+      "timeout": 300
+    },
+    "openrouter": {
+      "enabled": false,
+      "base_url": "https://openrouter.ai/api/v1",
+      "api_key": "",
+      "timeout": 120
+    },
+    "openai": {
+      "enabled": false,
+      "base_url": "https://api.openai.com/v1",
+      "api_key": "",
+      "timeout": 120
+    },
+    "anthropic": {
+      "enabled": false,
+      "base_url": "https://api.anthropic.com/v1",
+      "api_key": "",
+      "timeout": 120
+    }
+  },
+  "disabled_models": [],
+  "cloud_models": [],
+  "timeouts": {
+    "global": 300,
+    "per_model": {}
+  }
+}