{"repo": "astar-x/ai-roller-skating-checkin-main", "files": {"cloudfunctions/hunyuanVisionService/README.md": {"20d5b4cea088d2a8b6320f8ea41675e1cf02f6cb": {"t": 1785768877.9321675, "n": 1300, "h": "771de50c7f59db44", "c": "# hunyuanVisionService\n\nCloud function bridge for Tencent Hunyuan `hunyuan-turbos-vision-video`.\n\n## Deploy\n\n1. Open `cloudfunctions/hunyuanVisionService` in WeChat DevTools.\n2. Right click the folder and choose:\n   - Upload and Deploy: Install dependencies in cloud\n3. Enable HTTP access for this function and copy the full HTTPS URL.\n\n## Environment variables\n\n- `HUNYUAN_API_KEY` (required): OpenAI-compatible API key (`sk-...`)\n- `HUNYUAN_MODEL` (optional, default `hunyuan-turbos-vision-video`)\n- `HUNYUAN_API_ENDPOINT` (optional, default `https://api.hunyuan.cloud.tencent.com/v1/chat/completions`)\n- `HUNYUAN_TIMEOUT_MS` (optional, default `30000`)\n- `HUNYUAN_MAX_TOKENS` (optional, default `1200`)\n- `HUNYUAN_TEMPERATURE` (optional, default `0.2`)\n- `HUNYUAN_VIDEO_FPS` (optional, default `2`)\n- `STUB_ON_ERROR` (optional, default `false`)\n\nDuring debugging, keep `STUB_ON_ERROR=false` so failures are visible.\n\n## Wire into existing chain\n\nSet in `keypointInfer`:\n\n- `KEYPOINT_REAL_ENABLED=true`\n- `KEYPOINT_REAL_ENDPOINT=<HTTP URL of hunyuanVisionService>`\n- `KEYPOINT_REAL_TIMEOUT_MS=20000`\n\nKeep in `skateActionAnalyze`:\n\n- `KEYPOINT_API_ENABLED=true`\n- `KEYPOINT_API_FUNCTION=keypointInfer`\n\n## Verify\n\nAfter student analysis call:\n\n- `inferenceMode=api_keypoint`\n- `apiError` is empty\n\n"}}}}