Release v1.1.1: multi-provider UX, mixed routing, docs, and tests

Author: TrentPierce

.github/workflows/build-cross-platform.yml

@@ -0,0 +1,70 @@
+name: Build PolyCouncil (Cross-Platform)
+
+on:
+  workflow_dispatch:
+  push:
+    branches: [ main ]
+
+jobs:
+  build:
+    strategy:
+      fail-fast: false
+      matrix:
+        os: [windows-latest, ubuntu-latest, macos-latest]
+    runs-on: ${{ matrix.os }}
+
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+
+      - name: Setup Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: '3.11'
+
+      - name: Install dependencies
+        run: |
+          python -m pip install --upgrade pip
+          pip install -r requirements.txt
+          pip install pyinstaller
+
+      - name: Build (Windows)
+        if: runner.os == 'Windows'
+        shell: pwsh
+        run: |
+          python -m PyInstaller build_exe.spec --clean
+
+      - name: Build (Linux/macOS)
+        if: runner.os != 'Windows'
+        run: |
+          pyinstaller council.py --name PolyCouncil --clean --noconfirm --windowed --onefile
+
+      - name: Package artifact (Windows)
+        if: runner.os == 'Windows'
+        shell: pwsh
+        run: |
+          Compress-Archive -Path dist/PolyCouncil.exe -DestinationPath dist/PolyCouncil-windows.zip -Force
+
+      - name: Package artifact (Linux)
+        if: runner.os == 'Linux'
+        run: |
+          chmod +x dist/PolyCouncil || true
+          tar -czf dist/PolyCouncil-linux.tar.gz -C dist PolyCouncil
+
+      - name: Package artifact (macOS)
+        if: runner.os == 'macOS'
+        run: |
+          chmod +x dist/PolyCouncil || true
+          tar -czf dist/PolyCouncil-macos.tar.gz -C dist PolyCouncil
+
+      - name: Upload artifact
+        uses: actions/upload-artifact@v4
+        with:
+          name: PolyCouncil-${{ runner.os }}
+          path: |
+            dist/PolyCouncil.exe
+            dist/PolyCouncil-windows.zip
+            dist/PolyCouncil
+            dist/PolyCouncil-linux.tar.gz
+            dist/PolyCouncil-macos.tar.gz
+          if-no-files-found: ignore

.gitignore

@@ -21,6 +21,7 @@ token.json
 logs/
 *.log
 *.log.*
+polycouncil_crash.log
 
 # Models / weights / large artifacts
 models/
@@ -52,6 +53,8 @@ temp/
 *.tmp
 *.swp
 *.swo
+.pytest_cache/
+pytest-cache-files-*
 
 # Editors / workspace
 *.code-workspace
@@ -83,6 +86,7 @@ test_*.py
 # Icons and resources (keep only essential)
 *.ico
 *.png
+!docs/*.png
 *.rc
 PolyCouncil.rc
 
@@ -95,4 +99,4 @@ README_ICON_FIX.md
 *.sqlite
 *.sqlite3
 *.db-journal
-council_stats.db
+council_stats.db

README.md

@@ -2,7 +2,7 @@
 
 <div align="center">
 
-**A multi-model deliberation engine for LM Studio**
+**A multi-model deliberation engine for local and hosted providers**
 
 *Running a single model gives you one perspective. Running a council gives you… perspective squared.*
 
@@ -47,6 +47,10 @@ This creates a more reliable, research-friendly way to:
 - **Rubric-Based Scoring**: Each model evaluates every other response across customizable criteria (accuracy, clarity, completeness, etc.)
 - **Consensus Voting System**: Final answer chosen through weighted voting based on rubric scores
 - **Leaderboard Tracking**: Track which models perform best over time with persistent statistics
+- **Multi-Provider Support**:
+  - LM Studio (default, OpenAI-compatible)
+  - OpenAI-compatible APIs (custom base URL + API key)
+  - Ollama (local runtime)
 
 ### Persona System
 
@@ -69,7 +73,7 @@ This creates a more reliable, research-friendly way to:
 
 ### User Interface
 
-- **Centralized Settings**: Easy-to-use settings dialog for configuration
+- **Provider Profiles**: Save multiple providers and load/add models from each without retyping settings
 - **Debug Log Dock**: Dockable panel for detailed analysis of the deliberation process
 - **Modern UI**: Clean, intuitive interface built with PySide6 with automatic dark/light theme support
 - **Persistent Settings**: All preferences saved automatically
@@ -127,11 +131,15 @@ python council.py
 
 ### First Time Setup
 
-1. **Launch LM Studio** and ensure at least one model is loaded
+1. **Launch at least one model provider** (LM Studio, Ollama, or hosted API)
 2. **Start PolyCouncil** (either EXE or `python council.py`)
-3. **Configure Base URL** (default: `http://localhost:1234`) if LM Studio uses a different port
-4. **Click "Connect"** to discover available models
-5. **Select models** by checking the boxes next to model names
+3. **Choose Provider** in the top bar:
+   - Local: `LM Studio` (`http://localhost:1234`) or `Ollama` (`http://localhost:11434`)
+   - Hosted: `OpenAI-compatible API` + API service preset (`OpenAI`, `OpenRouter`, `Google Gemini`) + API key
+4. Click **Add Models** to append that provider's models
+5. Repeat step 3-4 for any additional providers
+6. Use **Add Provider** to save reusable provider configurations
+7. Select models (including mixed providers) and run the council
 
 ### Using Personas
 
@@ -203,13 +211,16 @@ Test how different personas affect model responses. Useful for understanding bia
 
 ![PolyCouncil Result](Polycouncil2.png)
 
+![Saved Providers Workflow](docs/saved-providers.png)
+
 ---
 
 ## Configuration
 
 PolyCouncil automatically saves your preferences, including:
-- LM Studio base URL
-- Selected models
+- Active provider selection and API service
+- API endpoint settings and keys
+- Saved provider profiles
 - Persona assignments
 - Concurrency settings
 - Debug mode preference
@@ -222,10 +233,10 @@ Settings are stored in `council_settings.json` in the application directory.
 ## Troubleshooting
 
 ### Models Not Appearing
-- Ensure LM Studio is running
-- Verify at least one model is loaded in LM Studio
-- Check that the base URL is correct (default: `http://localhost:1234`)
-- Click "Refresh Models" if needed
+- Verify the selected provider and endpoint are correct
+- For hosted APIs, confirm a valid API key is set
+- For LM Studio/Ollama, ensure the local server is running and models are loaded
+- Use **Add Models** to append, or **Replace List** to rebuild from one provider
 
 ### Slow Performance
 - Reduce "Max concurrent jobs" to 1-2

RELEASE_NOTES_v1.1.1.md

@@ -0,0 +1,46 @@
+# PolyCouncil v1.1.1 Release Notes
+
+**Release Date:** February 18, 2026
+
+## Highlights
+
+- Added first-class multi-provider support:
+  - LM Studio
+  - Ollama
+  - OpenAI-compatible hosted APIs
+- Added hosted API presets:
+  - OpenAI
+  - OpenRouter
+  - Google Gemini (OpenAI-compatible endpoint)
+- Added mixed-provider councils: local and hosted models can now run in the same round.
+
+## UX and Workflow Improvements
+
+- Provider setup is now centered in the main screen for faster use.
+- Added additive model loading workflow:
+  - **Add Models** appends provider models to the current list.
+  - **Replace List** explicitly rebuilds the list from one provider.
+  - **Clear Model List** removes all loaded models.
+- Added saved provider profiles:
+  - **Add Provider**, then **Use** or **Add Models** per saved provider.
+- Single-voter and selection workflows now work with provider-tagged model names.
+
+## Stability and Reliability Fixes
+
+- Fixed incorrect LM Studio model endpoint usage (`/v1/models` enforced for LM Studio).
+- Hardened provider switching and model list merge paths.
+- Added hosted-provider key validation before model discovery.
+- Added crash logging (`polycouncil_crash.log`) for unhandled exceptions.
+- Fixed discussion mode argument mismatch and added cancellation handling.
+- Fixed discussion export question capture.
+- Reduced capability detection overhead with metadata caching.
+- Replaced mutable default list arguments in runtime paths.
+
+## Developer and QA
+
+- Added automated tests for:
+  - Provider config/default behavior and endpoint construction
+  - Capability detection helpers
+  - Mixed-provider vote routing
+  - Voting JSON parsing/validation
+  - Discussion manager provider/cancellation behavior