mcp-tts-copilot-summary

# MCP TTS Copilot Summary Python MCP server for Windows text-to-speech that reads existing Copilot summaries using SAPI voice fallback. ## Behavior Summary - Exposes two MCP tools: - `speak_text` - `speak_existing_summary` - Consumes already-prepared summary text (no extra summarization pipeline). - Trigger strategy is MCP-only and instruction-driven. - Voice fallback policy: 1. Use Zira if available. 2. Else use first available SAPI voice. 3. If no SAPI voice is available, return non-fatal `skipped` with warning. - Speech call behavior is non-blocking by default (`MCP_TTS_SUMMARY_ASYNC_SPEECH=true`), so Copilot can continue responding immediately. ## Install ### pip ```powershell python -m pip install -e . ``` ### uv ```powershell uv pip install -e . ``` ## Run ```powershell python -m mcp_tts_copilot_summary.main ``` Uses stdio transport. ## Configuration Environment variables: - `MCP_TTS_SUMMARY_ENABLED` (default: `true`) - `MCP_TTS_SUMMARY_MIN_PARAGRAPHS` (default: `4`) - `MCP_TTS_SUMMARY_MIN_NEW_CODE_LINES` (default: `100`) - `MCP_TTS_SUMMARY_RATE` (default: `0`) - `MCP_TTS_SUMMARY_VOLUME` (default: `100`) - `MCP_TTS_SUMMARY_ASYNC_SPEECH` (default: `true`) Guard criteria used by tools: - Speak only when `paragraph_count > 4` OR `new_code_lines > 100`. - If metadata is not provided, tools proceed unless disabled. ## VS Code MCP Registration (Example) See [examples/mcp.json](examples/mcp.json) for a complete sample. If `mcp-tts-copilot-summary` is not on `PATH`, prefer launching with `python -m mcp_tts_copilot_summary.main`. For this workspace on Windows, use: ```json { "mcpServers": { "tts-copilot-summary": { "command": "python", "args": ["-m", "mcp_tts_copilot_summary.main"], "env": { "MCP_TTS_SUMMARY_ENABLED": "true", "MCP_TTS_SUMMARY_MIN_PARAGRAPHS": "4", "MCP_TTS_SUMMARY_MIN_NEW_CODE_LINES": "100", "MCP_TTS_SUMMARY_RATE": "0", "MCP_TTS_SUMMARY_VOLUME": "100" }…