[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M78ASMSFYFYECRA9CK327NXA":3,"skill-same:01M78ASMSFYFYECRA9CK327NXA":115},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":33,"commitSha":34,"skillDirectory":35,"forksCount":14,"repositoryHtmlUrl":36,"repositoryHomepageUrl":6,"ownerHtmlUrl":37,"ownerBlogUrl":38,"ownerDescription":39,"updatedAt":40,"install":41,"resources":54},"01M78ASMSFYFYECRA9CK327NXA","tao-train-nvpanoptix3d",null,"NVPanoptix3D for panoptic 3D scene reconstruction from posed RGB images. Produces 3D panoptic segmentation (semantic, instance, and panoptic masks) with occupancy completion. Built on a VGGT backbone with a Mask2Former-style head and 3D frustum reconstruction. Use when training, evaluating, exporting, or running inference for a TAO NVPanoptix3D model. Trigger phrases include \"train NVPanoptix3D\", \"panoptic 3D reconstruction\", \"3D scene segmentation\", \"occupancy completion\".","Solizardking\u002Ftrain2earn","Solizardking","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F108096810?v=4","01M786KDB10AYV042N0F9BD5B5","Clawd","01M786KDC5C3NPE4W44C5641EZ",0,"Apache-2.0","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fskills\u002Fnvidia\u002Ftao-train-nvpanoptix3d\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: tao-train-nvpanoptix3d\ndescription: NVPanoptix3D for panoptic 3D scene reconstruction from posed RGB images. Produces 3D panoptic segmentation\n  (semantic, instance, and panoptic masks) with occupancy completion. Built on a VGGT backbone with a Mask2Former-style head\n  and 3D frustum reconstruction. Use when training, evaluating, exporting, or running inference for a TAO NVPanoptix3D model.\n  Trigger phrases include \"train NVPanoptix3D\", \"panoptic 3D reconstruction\", \"3D scene segmentation\", \"occupancy completion\".\nlicense: Apache-2.0\ncompatibility: Requires docker + nvidia-container-toolkit.\nmetadata:\n  version: \"0.1.0\"\n  author: NVIDIA Corporation\nallowed-tools: Read Bash\ntags:\n- panoptic\n- 3d\n- reconstruction\n---\n\n# NVPanoptix3D\n\nNVPanoptix3D for panoptic 3D scene reconstruction from posed RGB images. Produces 3D panoptic segmentation (semantic, instance, and panoptic masks) with occupancy completion. Built on VGGT backbone with Mask2Former-style head and 3D frustum reconstruction.\n\nUses 2D and 3D stage checkpoints. Set train.checkpoint_2d and train.checkpoint_3d for staged initialization.\n\n## Dataclass Schemas\n\nGenerated TAO Core schemas are packaged in `schemas\u002F\u003Caction>.schema.json`, with `schemas\u002Fmanifest.json` listing available actions. Each generated schema also emits `references\u002Fspec_template_\u003Caction>.yaml` from the schema top-level `default` field. AutoML enablement is declared at the model layer in `references\u002Fskill_info.yaml` via `automl_enabled`. Runnable AutoML still requires `schemas\u002Ftrain.schema.json` and `references\u002Fspec_template_train.yaml` to exist and parse. Use the packaged train schema for `automl_default_parameters`, `automl_disabled_parameters`, defaults, min\u002Fmax bounds, enums, option weights, math conditions, dependencies, and popular parameters. Do not expect `~\u002Ftao-core` at runtime; maintainers regenerate schemas\u002Ftemplates before packaging the skill bank.\n\n## Train Action Policy\n\nThis model is AutoML-enabled at the model layer. Before handling any train-stage request, read `references\u002Fskill_info.yaml` and resolve the run override from either an explicit `automl_policy` value or the user's workflow request. Use `automl_policy: on` by default and only expose `on` \u002F `off` in new launch prompts. Treat phrases like \"turn off AutoML\", \"disable AutoML\", \"no HPO\", or \"plain training\" as `automl_policy: off` for this run only. When `automl_policy: on`, `automl_enabled: true`, and both `schemas\u002Ftrain.schema.json` and `references\u002Fspec_template_train.yaml` are packaged, route the train action through `tao-skill-bank:tao-run-automl` by default with this model's `skill_dir`. Preserve workflow\u002Fapplication overrides for datasets, specs, output directories, GPU\u002Fplatform settings, parent checkpoints, and `automl_policy`. Use direct model training only when `automl_policy: off` or the packaged train schema\u002Ftemplate is missing; in the missing-schema case, report that AutoML is enabled but not runnable for this model until schemas are generated.\n\nFor AutoML, use `train_loss` as the optimization metric with\n`direction=minimize`, and set `train.optim.monitor_name: train_loss` in\n`spec_overrides`. NVPanoptix3D train jobs emit `PRQ`, `RSQ`, and `RRQ` in\n`status.json`, and the training progress log emits `train_loss`; short or\nminimal jobs may not emit `val_loss`, including full-trial smoke runs.\nMulti-fidelity AutoML algorithms such as Hyperband, ASHA, and BOHB may promote\na checkpoint to a resume job that completes without emitting a fresh\n`train_loss` line. In that case, the AutoML metric is the carried-forward\nmetric from the source rung job that emitted `train_loss`; still verify the\npromoted job resumed from the explicit epoch\u002Fstep checkpoint, produced a real\ncheckpoint, and is usable for evaluate\u002Finference.\nNon-train actions such as `evaluate`, `inference`, `export`, and deploy flows stay in this model skill. The per-run `automl_policy` override does not change model metadata.\n\n## Training Requirements\n\n- **Dataset type:** nvpanoptix3d\n- **Formats:** front3d, matterport\n- **Monitoring metric:** train_loss\n- **AutoML direction:** minimize\n- For AutoML train jobs, use `train_loss`. For multi-fidelity resume jobs that do not emit a fresh `train_loss`,\n  compare AutoML's carried metric to the source rung job that emitted it.\n  Validation status KPIs are `PRQ`, `RSQ`, and `RRQ`; do not use `val_loss`\n  unless a specific run is known to emit it.\n\n### Per-Action Dataset Requirements\n\n| Action | Spec Key | Source | Files | List? |\n|---|---|---|---|---|\n| evaluate | dataset.frustum_mask_path | eval_dataset | meta\u002Ffrustum_mask.npz | No |\n| evaluate | dataset.label_map | eval_dataset | meta\u002Fcolormap.json | No |\n| evaluate | dataset.val.json_path | eval_dataset | meta\u002Fval.json | No |\n| evaluate | dataset.val.base_dir | eval_dataset |  | No |\n| evaluate | dataset.test.json_path | inference_dataset | meta\u002Ftest.json | No |\n| evaluate | dataset.test.base_dir | inference_dataset |  | No |\n| inference | dataset.frustum_mask_path | inference_dataset | meta\u002Ffrustum_mask.npz | No |\n| inference | dataset.label_map | inference_dataset | meta\u002Fcolormap.json | No |\n| inference | inference.images_dir | inference_dataset | flat folder of `.jpg`\u002F`.png` RGB images | No |\n| train | dataset.frustum_mask_path | train_datasets | meta\u002Ffrustum_mask.npz | No |\n| train | dataset.label_map | train_datasets | meta\u002Fcolormap.json | No |\n| train | dataset.train.json_path | train_datasets | meta\u002Ftrain.json | No |\n| train | dataset.train.base_dir | train_datasets |  | No |\n| train | dataset.val.json_path | eval_dataset | meta\u002Fval.json | No |\n| train | dataset.val.base_dir | eval_dataset |  | No |\n| train | dataset.test.json_path | inference_dataset | meta\u002Ftest.json | No |\n| train | dataset.test.base_dir | inference_dataset |  | No |\n\n### Typical Spec Overrides\n\nData source overrides are **mandatory for every action** — the agent MUST construct data source paths from the Per-Action Dataset Requirements table above and include them in `spec_overrides`.\nFor packaged S3 folders that store scene data as `data\u002Fimages.tar.gz`, the\nskill metadata requests extraction into the parent `data\u002F` directory because\nthe TAO loader expects `base_dir\u002Fdata\u002F\u003Cscene_id>\u002F...`.\n\n```python\nS3_TRAIN = \"s3:\u002F\u002Fbucket\u002Fdata\u002Ftrain\"\nS3_EVAL = \"s3:\u002F\u002Fbucket\u002Fdata\u002Feval\"\n```\n\n**train (mandatory data sources):**\n```python\n{\n    \"train.num_epochs\": 10,\n    \"train.checkpoint_interval\": 10,\n    \"train.validation_interval\": 10,\n    \"train.num_gpus\": 1,\n    \"dataset.enable_3d\": True,\n    \"dataset.contiguous_id\": True,\n    \"model.sem_seg_head.num_classes\": 13,\n    \"dataset.frustum_mask_path\": f\"{S3_TRAIN}\u002Fmeta\u002Ffrustum_mask.npz\",\n    \"dataset.label_map\": f\"{S3_TRAIN}\u002Fmeta\u002Fcolormap.json\",\n    \"dataset.train.json_path\": f\"{S3_TRAIN}\u002Fmeta\u002Ftrain.json\",\n    \"dataset.train.base_dir\": f\"{S3_TRAIN}\",\n    \"dataset.val.json_path\": f\"{S3_EVAL}\u002Fmeta\u002Fval.json\",\n    \"dataset.val.base_dir\": f\"{S3_EVAL}\",\n    \"dataset.test.json_path\": f\"{S3_EVAL}\u002Fmeta\u002Ftest.json\",\n    \"dataset.test.base_dir\": f\"{S3_EVAL}\",\n}\n```\n\n**evaluate (mandatory data sources):**\n```python\n{\n    \"evaluate.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"dataset.enable_3d\": True,\n    \"dataset.contiguous_id\": True,\n    \"dataset.frustum_mask_path\": f\"{S3_EVAL}\u002Fmeta\u002Ffrustum_mask.npz\",\n    \"dataset.label_map\": f\"{S3_EVAL}\u002Fmeta\u002Fcolormap.json\",\n    \"dataset.val.json_path\": f\"{S3_EVAL}\u002Fmeta\u002Fval.json\",\n    \"dataset.val.base_dir\": f\"{S3_EVAL}\",\n    \"dataset.test.json_path\": f\"{S3_EVAL}\u002Fmeta\u002Ftest.json\",\n    \"dataset.test.base_dir\": f\"{S3_EVAL}\",\n}\n```\n\n**inference (mandatory data sources):**\n```python\n{\n    \"inference.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"dataset.enable_3d\": True,\n    \"dataset.frustum_mask_path\": f\"{S3_EVAL}\u002Fmeta\u002Ffrustum_mask.npz\",\n    \"dataset.label_map\": f\"{S3_EVAL}\u002Fmeta\u002Fcolormap.json\",\n    \"inference.images_dir\": \"\u002Fpath\u002Fto\u002Fflat_rgb_images\",\n}\n```\n## Eval Dataset\n\nOptional. Val\u002Ftest splits configured via dataset.val and dataset.test paths.\n\n## Important Parameters\n\n- **model.sem_seg_head.num_classes**: Number of semantic classes. Default 13.\n- **model.mode**: Prediction mode. Options: panoptic, instance, semantic. Default panoptic.\n- **model.backbone_type**: Backbone. Default vggt (only option in schema).\n- **model.mask_former.num_object_queries**: Object queries. Default 100.\n- **model.mask_former.dec_layers**: Decoder layers. Default 10.\n- **model.frustum3d.truncation**: 3D frustum truncation. Default 3.\n- **model.frustum3d.panoptic_weight**: Panoptic loss weight. Default 25.\n- **model.frustum3d.completion_weights**: Completion loss weights. Default [50, 25, 10].\n- **dataset.name**: Dataset name. Options: front3d, matterport, synthetic_hospital, synthetic_warehouse.\n- **dataset.contiguous_id**: Set `True` when the label-map JSON already\n  supplies `trainId` values for its category IDs; leaving the default `False`\n  can synthesize placeholder categories without `trainId` and fail during\n  metadata construction.\n- **dataset.downsample_factor**: Image downsample factor. Default 1 (Front3D), 2 (Matterport).\n- **dataset.target_size**: Target image size. Default [320, 240].\n- **dataset.depth_min**: Min depth. Default 0.4 meters.\n- **dataset.depth_max**: Max depth. Default 6.0 meters.\n- **train.lr**: Learning rate. Default 2e-4. backbone_multiplier=0.1.\n- **train.lr_scheduler**: Options: MultiStep, Warmuppoly. Milestones [88, 96].\n- **train.precision**: Only `fp32` is supported by the current train code.\n- **train.distributed_strategy**: Options: ddp, fsdp. activation_checkpoint=True by default.\n- **train.clip_grad_norm**: Gradient clipping norm. Default 0.1.\n- **export.onnx_file_2d**: ONNX path for 2D model component.\n- **export.max_voxels**: Max voxels for engine input. Default 700000.\n- **inference.mode**: Options: semantic, instance, panoptic.\n\n## Multi-GPU \u002F Multi-Node\n\n**Launch method:** Lightning-managed (single `python` process, Lightning spawns workers).\n\n| Spec Key | Description | Default |\n|----------|-------------|---------|\n| `train.num_gpus` | Number of GPUs | 1 |\n| `train.gpu_ids` | GPU device indices | [0] |\n| `train.num_nodes` | Number of nodes | 1 |\n| `train.distributed_strategy` | `ddp` only | `ddp` |\n\n- **`fsdp` is NOT supported** for NVPanoptix3D (code only handles `ddp`)\n- `ddp` with activation checkpointing (enabled by default): `find_unused_parameters=False`\n- `ddp` without: `find_unused_parameters=True`\n- FAN backbones with 3D enabled auto-enable `sync_batchnorm`\n\n**Multi-node env vars** (set by orchestrator): `WORLD_SIZE`, `NODE_RANK`, `MASTER_ADDR`, `MASTER_PORT`, `NUM_GPU_PER_NODE`.\n\n## Export \u002F TRT Defaults\n\n- Exports the 2D ONNX model to `export.onnx_file_2d`. The current export\n  entrypoint calls `export_2d_model`; `export.onnx_file_3d` is present in the\n  schema but not produced by this toolkit image.\n- TRT data types: FP32, FP16 only\n- max_voxels: 700000 (engine input tensor limit)\n\n## Hardware\n\nMinimum 2 GPU(s), recommended 4 GPU(s). 40GB+ (A100 recommended) VRAM per GPU. 3D reconstruction is very memory intensive. Use `train.precision: fp32`; the current training entrypoint rejects fp16. activation_checkpoint enabled by default. FSDP for multi-node. AutoML is enabled at the model layer; preserve this GPU\u002FVRAM guidance when routing train through AutoML.\n\n## Error Patterns\n\n**`nvpanoptix3d: not found` in the PyTorch image**: Use the packaged module\nentrypoint command:\n`python -m nvidia_tao_pytorch.cv.nvpanoptix3d.entrypoint.nvpanoptix3d \u003Caction> -e \u003Cspec>`.\nThe 7.0 PyTorch image contains the NVPanoptix3D package but does not expose a\n`nvpanoptix3d` console script.\n\n**Missing frustum mask**: Ensure meta\u002Ffrustum_mask.npz is present in the dataset directory.\n\n**Downsample factor mismatch**: Use downsample_factor=2 for Matterport3D, 1 for Front3D \u002F synthetic datasets.\n\n**3D occupancy OOM**: Reduce frustum_dims or grid_dimensions if running out of GPU memory during 3D reconstruction.\n\n**fp16 precision rejected**: The schema advertises `fp16`, but the current\ntraining entrypoint raises `ValueError: Only fp32 precision is supported.` Use\n`train.precision: fp32` for train and resume\u002Fretrain.\n\n**Inference dataloader length is zero**: `inference.images_dir` is scanned only\nfor top-level `.jpg` and `.png` files. If the S3 test archive extracts to\nscene subdirectories, create or point to a flat folder of real RGB images before\nrunning inference.\n\n**3D ONNX missing after export**: The current export entrypoint only calls the\n2D ONNX exporter and writes `export.onnx_file_2d`. Do not require\n`export.onnx_file_3d` unless the toolkit image adds a 3D exporter.\n\n**Resume stops at the epoch boundary**: A one-epoch smoke run writes an\nend-of-epoch checkpoint such as `model_epoch_000_step_00020.pth`. Resuming with\n`train.num_epochs` set only one epoch beyond the original run can restore the\ncheckpoint and stop without producing a new epoch checkpoint. When validating\nactual retraining from an epoch-boundary checkpoint, set `train.num_epochs` at\nleast two epochs beyond the source smoke run and raise `train.optim.max_steps`\naccordingly. For example, resuming from `model_epoch_000_step_00020.pth` needs\n`train.num_epochs: 3` and enough max steps to produce a new exact epoch\u002Fstep\ncheckpoint such as `model_epoch_001_step_00040.pth` before handing the model to\nevaluate, inference, or export.\n\n## Spec Param \u002F Parent Model Inference\n\nModel-specific inference mappings belong in this MD file, not in `config.json`. Generated runners should read this section and apply the mappings with SDK helpers before `create_job()`. This mirrors the old microservices `infer_params.py` flow.\n\nModel-specific handoff mappings:\n\n| Action | Spec Field | Inference Function | Meaning |\n|---|---|---|---|\n| evaluate | `encryption_key` | `key` | encryption key |\n| evaluate | `evaluate.checkpoint` | `parent_model` | model file inferred from the parent job results folder |\n| evaluate | `results_dir` | `output_dir` | current job results directory |\n| export | `encryption_key` | `key` | encryption key |\n| export | `export.checkpoint` | `parent_model` | model file inferred from the parent job results folder |\n| export | `export.onnx_file_2d` | `create_onnx_file_2d` | output 2D ONNX path |\n| export | `results_dir` | `output_dir` | current job results directory |\n| inference | `encryption_key` | `key` | encryption key |\n| inference | `inference.checkpoint` | `parent_model` | model file inferred from the parent job results folder |\n| inference | `results_dir` | `output_dir` | current job results directory |\n| train | `encryption_key` | `key` | encryption key |\n| train | `results_dir` | `output_dir` | current job results directory |\n| train | `train.checkpoint_2d` | `parent_model_or_ptm` | parent model if available, otherwise PTM |\n| train | `train.checkpoint_3d` | `ptm` | pretrained model |\n| train | `train.resume_training_checkpoint_path` | `resume_model` | model file inferred from the current job results folder |\n\nFor `parent_model` or `parent_model_folder`, pass the upstream train\u002Fexport\u002FAutoML child job id as `parent_job_id`. The SDK lists the parent result folder, filters checkpoint artifacts, and returns the selected model file or folder. Do not add these mappings back to `config.json` and do not patch generated runner scripts to guess checkpoint paths.\n","# NVPanoptix3D\n\nNVPanoptix3D for panoptic 3D scene reconstruction from posed RGB images. Produces 3D panoptic segmentation (semantic, instance, and panoptic masks) with occupancy completion. Built on VGGT backbone with Mask2Former-style head and 3D frustum reconstruction.\n\nUses 2D and 3D stage checkpoints. Set train.checkpoint_2d and train.checkpoint_3d for staged initialization.\n\n## Dataclass Schemas\n\nGenerated TAO Core schemas are packaged in `schemas\u002F\u003Caction>.schema.json`, with `schemas\u002Fmanifest.json` listing available actions. Each generated schema also emits `references\u002Fspec_template_\u003Caction>.yaml` from the schema top-level `default` field. AutoML enablement is declared at the model layer in `references\u002Fskill_info.yaml` via `automl_enabled`. Runnable AutoML still requires `schemas\u002Ftrain.schema.json` and `references\u002Fspec_template_train.yaml` to exist and parse. Use the packaged train schema for `automl_default_parameters`, `automl_disabled_parameters`, defaults, min\u002Fmax bounds, enums, option weights, math conditions, dependencies, and popular parameters. Do not expect `~\u002Ftao-core` at runtime; maintainers regenerate schemas\u002Ftemplates before packaging the skill bank.\n\n## Train Action Policy\n\nThis model is AutoML-enabled at the model layer. Before handling any train-stage request, read `references\u002Fskill_info.yaml` and resolve the run override from either an explicit `automl_policy` value or the user's workflow request. Use `automl_policy: on` by default and only expose `on` \u002F `off` in new launch prompts. Treat phrases like \"turn off AutoML\", \"disable AutoML\", \"no HPO\", or \"plain training\" as `automl_policy: off` for this run only. When `automl_policy: on`, `automl_enabled: true`, and both `schemas\u002Ftrain.schema.json` and `references\u002Fspec_template_train.yaml` are packaged, route the train action through `tao-skill-bank:tao-run-automl` by default with this model's `skill_dir`. Preserve workflow\u002Fapplication overrides for datasets, specs, output directories, GPU\u002Fplatform settings, parent checkpoints, and `automl_policy`. Use direct model training only when `automl_policy: off` or the packaged train schema\u002Ftemplate is missing; in the missing-schema case, report that AutoML is enabled but not runnable for this model until schemas are generated.\n\nFor AutoML, use `train_loss` as the optimization metric with\n`direction=minimize`, and set `train.optim.monitor_name: train_loss` in\n`spec_overrides`. NVPanoptix3D train jobs emit `PRQ`, `RSQ`, and `RRQ` in\n`status.json`, and the training progress log emits `train_loss`; short or\nminimal jobs may not emit `val_loss`, including full-trial smoke runs.\nMulti-fidelity AutoML algorithms such as Hyperband, ASHA, and BOHB may promote\na checkpoint to a resume job that completes without emitting a fresh\n`train_loss` line. In that case, the AutoML metric is the carried-forward\nmetric from the source rung job that emitted `train_loss`; still verify the\npromoted job resumed from the explicit epoch\u002Fstep checkpoint, produced a real\ncheckpoint, and is usable for evaluate\u002Finference.\nNon-train actions such as `evaluate`, `inference`, `export`, and deploy flows stay in this model skill. The per-run `automl_policy` override does not change model metadata.\n\n## Training Requirements\n\n- **Dataset type:** nvpanoptix3d\n- **Formats:** front3d, matterport\n- **Monitoring metric:** train_loss\n- **AutoML direction:** minimize\n- For AutoML train jobs, use `train_loss`. For multi-fidelity resume jobs that do not emit a fresh `train_loss`,\n  compare AutoML's carried metric to the source rung job that emitted it.\n  Validation status KPIs are `PRQ`, `RSQ`, and `RRQ`; do not use `val_loss`\n  unless a specific run is known to emit it.\n\n### Per-Action Dataset Requirements\n\n| Action | Spec Key | Source | Files | List? |\n|---|---|---|---|---|\n| evaluate | dataset.frustum_mask_path | eval_dataset | meta\u002Ffrustum_mask.npz | No |\n| evaluate | dataset.label_map | eval_dataset | meta\u002Fcolormap.json | No |\n| evaluate | dataset.val.json_path | eval_dataset | meta\u002Fval.json | No |\n| evaluate | dataset.val.base_dir | eval_dataset |  | No |\n| evaluate | dataset.test.json_path | inference_dataset | meta\u002Ftest.json | No |\n| evaluate | dataset.test.base_dir | inference_dataset |  | No |\n| inference | dataset.frustum_mask_path | inference_dataset | meta\u002Ffrustum_mask.npz | No |\n| inference | dataset.label_map | inference_dataset | meta\u002Fcolormap.json | No |\n| inference | inference.images_dir | inference_dataset | flat folder of `.jpg`\u002F`.png` RGB images | No |\n| train | dataset.frustum_mask_path | train_datasets | meta\u002Ffrustum_mask.npz | No |\n| train | dataset.label_map | train_datasets | meta\u002Fcolormap.json | No |\n| train | dataset.train.json_path | train_datasets | meta\u002Ftrain.json | No |\n| train | dataset.train.base_dir | train_datasets |  | No |\n| train | dataset.val.json_path | eval_dataset | meta\u002Fval.json | No |\n| train | dataset.val.base_dir | eval_dataset |  | No |\n| train | dataset.test.json_path | inference_dataset | meta\u002Ftest.json | No |\n| train | dataset.test.base_dir | inference_dataset |  | No |\n\n### Typical Spec Overrides\n\nData source overrides are **mandatory for every action** — the agent MUST construct data source paths from the Per-Action Dataset Requirements table above and include them in `spec_overrides`.\nFor packaged S3 folders that store scene data as `data\u002Fimages.tar.gz`, the\nskill metadata requests extraction into the parent `data\u002F` directory because\nthe TAO loader expects `base_dir\u002Fdata\u002F\u003Cscene_id>\u002F...`.\n\n```python\nS3_TRAIN = \"s3:\u002F\u002Fbucket\u002Fdata\u002Ftrain\"\nS3_EVAL = \"s3:\u002F\u002Fbucket\u002Fdata\u002Feval\"\n```\n\n**train (mandatory data sources):**\n```python\n{\n    \"train.num_epochs\": 10,\n    \"train.checkpoint_interval\": 10,\n    \"train.validation_interval\": 10,\n    \"train.num_gpus\": 1,\n    \"dataset.enable_3d\": True,\n    \"dataset.contiguous_id\": True,\n    \"model.sem_seg_head.num_classes\": 13,\n    \"dataset.frustum_mask_path\": f\"{S3_TRAIN}\u002Fmeta\u002Ffrustum_mask.npz\",\n    \"dataset.label_map\": f\"{S3_TRAIN}\u002Fmeta\u002Fcolormap.json\",\n    \"dataset.train.json_path\": f\"{S3_TRAIN}\u002Fmeta\u002Ftrain.json\",\n    \"dataset.train.base_dir\": f\"{S3_TRAIN}\",\n    \"dataset.val.json_path\": f\"{S3_EVAL}\u002Fmeta\u002Fval.json\",\n    \"dataset.val.base_dir\": f\"{S3_EVAL}\",\n    \"dataset.test.json_path\": f\"{S3_EVAL}\u002Fmeta\u002Ftest.json\",\n    \"dataset.test.base_dir\": f\"{S3_EVAL}\",\n}\n```\n\n**evaluate (mandatory data sources):**\n```python\n{\n    \"evaluate.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"dataset.enable_3d\": True,\n    \"dataset.contiguous_id\": True,\n    \"dataset.frustum_mask_path\": f\"{S3_EVAL}\u002Fmeta\u002Ffrustum_mask.npz\",\n    \"dataset.label_map\": f\"{S3_EVAL}\u002Fmeta\u002Fcolormap.json\",\n    \"dataset.val.json_path\": f\"{S3_EVAL}\u002Fmeta\u002Fval.json\",\n    \"dataset.val.base_dir\": f\"{S3_EVAL}\",\n    \"dataset.test.json_path\": f\"{S3_EVAL}\u002Fmeta\u002Ftest.json\",\n    \"dataset.test.base_dir\": f\"{S3_EVAL}\",\n}\n```\n\n**inference (mandatory data sources):**\n```python\n{\n    \"inference.checkpoint\": \"\u003Cselected train\u002FAutoML checkpoint>\",\n    \"dataset.enable_3d\": True,\n    \"dataset.frustum_mask_path\": f\"{S3_EVAL}\u002Fmeta\u002Ffrustum_mask.npz\",\n    \"dataset.label_map\": f\"{S3_EVAL}\u002Fmeta\u002Fcolormap.json\",\n    \"inference.images_dir\": \"\u002Fpath\u002Fto\u002Fflat_rgb_images\",\n}\n```\n## Eval Dataset\n\nOptional. Val\u002Ftest splits configured via dataset.val and dataset.test paths.\n\n## Important Parameters\n\n- **model.sem_seg_head.num_classes**: Number of semantic classes. Default 13.\n- **model.mode**: Prediction mode. Options: panoptic, instance, semantic. Default panoptic.\n- **model.backbone_type**: Backbone. Default vggt (only option in schema).\n- **model.mask_former.num_object_queries**: Object queries. Default 100.\n- **model.mask_former.dec_layers**: Decoder layers. Default 10.\n- **model.frustum3d.truncation**: 3D frustum truncation. Default 3.\n- **model.frustum3d.panoptic_weight**: Panoptic loss weight. Default 25.\n- **model.frustum3d.completion_weights**: Completion loss weights. Default [50, 25, 10].\n- **dataset.name**: Dataset name. Options: front3d, matterport, synthetic_hospital, synthetic_warehouse.\n- **dataset.contiguous_id**: Set `True` when the label-map JSON already\n  supplies `trainId` values for its category IDs; leaving the default `False`\n  can synthesize placeholder categories without `trainId` and fail during\n  metadata construction.\n- **dataset.downsample_factor**: Image downsample factor. Default 1 (Front3D), 2 (Matterport).\n- **dataset.target_size**: Target image size. Default [320, 240].\n- **dataset.depth_min**: Min depth. Default 0.4 meters.\n- **dataset.depth_max**: Max depth. Default 6.0 meters.\n- **train.lr**: Learning rate. Default 2e-4. backbone_multiplier=0.1.\n- **train.lr_scheduler**: Options: MultiStep, Warmuppoly. Milestones [88, 96].\n- **train.precision**: Only `fp32` is supported by the current train code.\n- **train.distributed_strategy**: Options: ddp, fsdp. activation_checkpoint=True by default.\n- **train.clip_grad_norm**: Gradient clipping norm. Default 0.1.\n- **export.onnx_file_2d**: ONNX path for 2D model component.\n- **export.max_voxels**: Max voxels for engine input. Default 700000.\n- **inference.mode**: Options: semantic, instance, panoptic.\n\n## Multi-GPU \u002F Multi-Node\n\n**Launch method:** Lightning-managed (single `python` process, Lightning spawns workers).\n\n| Spec Key | Description | Default |\n|----------|-------------|---------|\n| `train.num_gpus` | Number of GPUs | 1 |\n| `train.gpu_ids` | GPU device indices | [0] |\n| `train.num_nodes` | Number of nodes | 1 |\n| `train.distributed_strategy` | `ddp` only | `ddp` |\n\n- **`fsdp` is NOT supported** for NVPanoptix3D (code only handles `ddp`)\n- `ddp` with activation checkpointing (enabled by default): `find_unused_parameters=False`\n- `ddp` without: `find_unused_parameters=True`\n- FAN backbones with 3D enabled auto-enable `sync_batchnorm`\n\n**Multi-node env vars** (set by orchestrator): `WORLD_SIZE`, `NODE_RANK`, `MASTER_ADDR`, `MASTER_PORT`, `NUM_GPU_PER_NODE`.\n\n## Export \u002F TRT Defaults\n\n- Exports the 2D ONNX model to `export.onnx_file_2d`. The current export\n  entrypoint calls `export_2d_model`; `export.onnx_file_3d` is present in the\n  schema but not produced by this toolkit image.\n- TRT data types: FP32, FP16 only\n- max_voxels: 700000 (engine input tensor limit)\n\n## Hardware\n\nMinimum 2 GPU(s), recommended 4 GPU(s). 40GB+ (A100 recommended) VRAM per GPU. 3D reconstruction is very memory intensive. Use `train.precision: fp32`; the current training entrypoint rejects fp16. activation_checkpoint enabled by default. FSDP for multi-node. AutoML is enabled at the model layer; preserve this GPU\u002FVRAM guidance when routing train through AutoML.\n\n## Error Patterns\n\n**`nvpanoptix3d: not found` in the PyTorch image**: Use the packaged module\nentrypoint command:\n`python -m nvidia_tao_pytorch.cv.nvpanoptix3d.entrypoint.nvpanoptix3d \u003Caction> -e \u003Cspec>`.\nThe 7.0 PyTorch image contains the NVPanoptix3D package but does not expose a\n`nvpanoptix3d` console script.\n\n**Missing frustum mask**: Ensure meta\u002Ffrustum_mask.npz is present in the dataset directory.\n\n**Downsample factor mismatch**: Use downsample_factor=2 for Matterport3D, 1 for Front3D \u002F synthetic datasets.\n\n**3D occupancy OOM**: Reduce frustum_dims or grid_dimensions if running out of GPU memory during 3D reconstruction.\n\n**fp16 precision rejected**: The schema advertises `fp16`, but the current\ntraining entrypoint raises `ValueError: Only fp32 precision is supported.` Use\n`train.precision: fp32` for train and resume\u002Fretrain.\n\n**Inference dataloader length is zero**: `inference.images_dir` is scanned only\nfor top-level `.jpg` and `.png` files. If the S3 test archive extracts to\nscene subdirectories, create or point to a flat folder of real RGB images before\nrunning inference.\n\n**3D ONNX missing after export**: The current export entrypoint only calls the\n2D ONNX exporter and writes `export.onnx_file_2d`. Do not require\n`export.onnx_file_3d` unless the toolkit image adds a 3D exporter.\n\n**Resume stops at the epoch boundary**: A one-epoch smoke run writes an\nend-of-epoch checkpoint such as `model_epoch_000_step_00020.pth`. Resuming with\n`train.num_epochs` set only one epoch beyond the original run can restore the\ncheckpoint and stop without producing a new epoch checkpoint. When validating\nactual retraining from an epoch-boundary checkpoint, set `train.num_epochs` at\nleast two epochs beyond the source smoke run and raise `train.optim.max_steps`\naccordingly. For example, resuming from `model_epoch_000_step_00020.pth` needs\n`train.num_epochs: 3` and enough max steps to produce a new exact epoch\u002Fstep\ncheckpoint such as `model_epoch_001_step_00040.pth` before handing the model to\nevaluate, inference, or export.\n\n## Spec Param \u002F Parent Model Inference\n\nModel-specific inference mappings belong in this MD file, not in `config.json`. Generated runners should read this section and apply the mappings with SDK helpers before `create_job()`. This mirrors the old microservices `infer_params.py` flow.\n\nModel-specific handoff mappings:\n\n| Action | Spec Field | Inference Function | Meaning |\n|---|---|---|---|\n| evaluate | `encryption_key` | `key` | encryption key |\n| evaluate | `evaluate.checkpoint` | `parent_model` | model file inferred from the parent job results folder |\n| evaluate | `results_dir` | `output_dir` | current job results directory |\n| export | `encryption_key` | `key` | encryption key |\n| export | `export.checkpoint` | `parent_model` | model file inferred from the parent job results folder |\n| export | `export.onnx_file_2d` | `create_onnx_file_2d` | output 2D ONNX path |\n| export | `results_dir` | `output_dir` | current job results directory |\n| inference | `encryption_key` | `key` | encryption key |\n| inference | `inference.checkpoint` | `parent_model` | model file inferred from the parent job results folder |\n| inference | `results_dir` | `output_dir` | current job results directory |\n| train | `encryption_key` | `key` | encryption key |\n| train | `results_dir` | `output_dir` | current job results directory |\n| train | `train.checkpoint_2d` | `parent_model_or_ptm` | parent model if available, otherwise PTM |\n| train | `train.checkpoint_3d` | `ptm` | pretrained model |\n| train | `train.resume_training_checkpoint_path` | `resume_model` | model file inferred from the current job results folder |\n\nFor `parent_model` or `parent_model_folder`, pass the upstream train\u002Fexport\u002FAutoML child job id as `parent_job_id`. The SDK lists the parent result folder, filters checkpoint artifacts, and returns the selected model file or folder. Do not add these mappings back to `config.json` and do not patch generated runner scripts to guess checkpoint paths.\n",{"name":5,"tags":24,"license":15,"metadata":28,"description":7,"allowed-tools":31,"compatibility":32},[25,26,27],"panoptic","3d","reconstruction",{"author":29,"version":30},"NVIDIA Corporation","0.1.0","Read Bash","Requires docker + nvidia-container-toolkit.","main","86cc27ffd16af6cb85948f1155578e6f3f61cc4c","skillhub\u002Fskills\u002Fnvidia\u002Ftao-train-nvpanoptix3d","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn","https:\u002F\u002Fgithub.com\u002FSolizardking","https:\u002F\u002Fmusebook.trade","Got a mile high and never came down","2026-10-08 11:52:53",{"source":8,"skillName":5,"commands":42,"prompt":52,"zipAvailable":53},[43,46,49],{"tool":44,"command":45},"npx","npx skills add Solizardking\u002Ftrain2earn --skill tao-train-nvpanoptix3d",{"tool":47,"command":48},"bunx","bunx skills add Solizardking\u002Ftrain2earn --skill tao-train-nvpanoptix3d",{"tool":50,"command":51},"pnpm","pnpm dlx skills add Solizardking\u002Ftrain2earn --skill tao-train-nvpanoptix3d","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fskills\u002Fnvidia\u002Ftao-train-nvpanoptix3d\u002FSKILL.md",true,[55,61,65,69,73,77,81,85,89,93,97,101,105,109],{"path":56,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":59,"sourceHtmlUrl":60},"BENCHMARK.md","REFERENCE","参考文档",4087,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fskills\u002Fnvidia\u002Ftao-train-nvpanoptix3d\u002FBENCHMARK.md",{"path":62,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":63,"sourceHtmlUrl":64},"evals\u002Fevals.json",824,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fskills\u002Fnvidia\u002Ftao-train-nvpanoptix3d\u002Fevals\u002Fevals.json",{"path":66,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":67,"sourceHtmlUrl":68},"references\u002Fskill_info.yaml",6388,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fskills\u002Fnvidia\u002Ftao-train-nvpanoptix3d\u002Freferences\u002Fskill_info.yaml",{"path":70,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":71,"sourceHtmlUrl":72},"references\u002Fspec_template_evaluate.yaml",3694,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fskills\u002Fnvidia\u002Ftao-train-nvpanoptix3d\u002Freferences\u002Fspec_template_evaluate.yaml",{"path":74,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":75,"sourceHtmlUrl":76},"references\u002Fspec_template_export.yaml",3844,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fskills\u002Fnvidia\u002Ftao-train-nvpanoptix3d\u002Freferences\u002Fspec_template_export.yaml",{"path":78,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":79,"sourceHtmlUrl":80},"references\u002Fspec_template_inference.yaml",3729,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fskills\u002Fnvidia\u002Ftao-train-nvpanoptix3d\u002Freferences\u002Fspec_template_inference.yaml",{"path":82,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":83,"sourceHtmlUrl":84},"references\u002Fspec_template_train.yaml",3569,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fskills\u002Fnvidia\u002Ftao-train-nvpanoptix3d\u002Freferences\u002Fspec_template_train.yaml",{"path":86,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":87,"sourceHtmlUrl":88},"schemas\u002Fevaluate.schema.json",59682,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fskills\u002Fnvidia\u002Ftao-train-nvpanoptix3d\u002Fschemas\u002Fevaluate.schema.json",{"path":90,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":91,"sourceHtmlUrl":92},"schemas\u002Fexport.schema.json",61373,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fskills\u002Fnvidia\u002Ftao-train-nvpanoptix3d\u002Fschemas\u002Fexport.schema.json",{"path":94,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":95,"sourceHtmlUrl":96},"schemas\u002Finference.schema.json",60142,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fskills\u002Fnvidia\u002Ftao-train-nvpanoptix3d\u002Fschemas\u002Finference.schema.json",{"path":98,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":99,"sourceHtmlUrl":100},"schemas\u002Fmanifest.json",10986,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fskills\u002Fnvidia\u002Ftao-train-nvpanoptix3d\u002Fschemas\u002Fmanifest.json",{"path":102,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":103,"sourceHtmlUrl":104},"schemas\u002Ftrain.schema.json",57012,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fskills\u002Fnvidia\u002Ftao-train-nvpanoptix3d\u002Fschemas\u002Ftrain.schema.json",{"path":106,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":107,"sourceHtmlUrl":108},"skill-card.md",4160,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fskills\u002Fnvidia\u002Ftao-train-nvpanoptix3d\u002Fskill-card.md",{"path":110,"resourceType":111,"resourceTypeDesc":112,"sizeBytes":113,"sourceHtmlUrl":114},"skill.oms.sig","OTHER","其他",7041,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fskills\u002Fnvidia\u002Ftao-train-nvpanoptix3d\u002Fskill.oms.sig",[116,124,132,140,147,155],{"skillNo":117,"name":118,"nameZh":6,"description":119,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":120,"licenseFileUrl":6,"sourceHtmlUrl":121,"categories":122,"keywords":123,"publishedAt":19},"01M78B94RSCAKMH1QS79VE6DG9","testing","For testing with Light Protocol programs and clients on localnet, devnet, and mainnet validation.","MIT","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fskills\u002Ftesting\u002FSKILL.md",[],[],{"skillNo":125,"name":126,"nameZh":6,"description":127,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":120,"licenseFileUrl":128,"sourceHtmlUrl":129,"categories":130,"keywords":131,"publishedAt":19},"01M78B81BKY8RX12HETPG8YKWK","qedgen","Formally verify programs by writing Lean 4 proofs. Trigger this skill whenever the user wants to formally verify code, generate Lean 4 proofs, prove properties about algorithms or smart contracts, verify invariants, convert program logic into formal specifications, or anything involving Lean 4 and formal verification. Also trigger when the user mentions \"qedgen\", \"lean proof\", \"formal proof\", \"verify my code\", \"prove correctness\", \"formal verification\", or wants mathematical guarantees about their implementation.","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fskills\u002Fsolana-formal-verification\u002FLICENSE","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fskills\u002Fsolana-formal-verification\u002FSKILL.md",[],[],{"skillNo":133,"name":134,"nameZh":6,"description":135,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":120,"licenseFileUrl":136,"sourceHtmlUrl":137,"categories":138,"keywords":139,"publishedAt":19},"01M78B7Q7C5K54E01DN83A60KG","solana-dev","Use when user asks to \"build a Solana dapp\", \"write an Anchor program\", \"create a token\", \"debug Solana errors\", \"set up wallet connection\", \"test my Solana program\", \"deploy to devnet\", or \"explain Solana concepts\" (rent, accounts, PDAs, CPIs, etc.). Also use for quick on-chain lookups via public RPC + curl — \"what's the balance of \u003Cwallet>\", \"look up transaction \u003Csig>\", \"token balance for \u003Caccount>\", \"check this address on mainnet\u002Fdevnet\". End-to-end Solana development playbook covering wallet connection, Anchor\u002FPinocchio programs, Codama client generation, LiteSVM\u002FMollusk\u002FSurfpool testing, security checklists, and JSON-RPC curl lookups against public clusters. Integrates with the Solana MCP server for live documentation search. Prefers framework-kit (@solana\u002Fclient + @solana\u002Freact-hooks) for UI, wallet-standard-first connection (incl. ConnectorKit), @solana\u002Fkit for client\u002FRPC code, and @solana\u002Fweb3-compat for legacy boundaries.","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fskills\u002Fsolana-dev\u002FLICENSE","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fskills\u002Fsolana-dev\u002FSKILL.md",[],[],{"skillNo":141,"name":142,"nameZh":6,"description":143,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":120,"licenseFileUrl":6,"sourceHtmlUrl":144,"categories":145,"keywords":146,"publishedAt":19},"01M78B7HXD6N6GWKSVRQ8F6Q31","solana-common-errors","Diagnose and fix common Solana development errors — GLIBC mismatches, Anchor version conflicts, cargo build-sbf \u002F platform-tools failures, LiteSVM issues, edition2024 crate pins, RPC airdrops, and Anchor 0.29→0.32 migrations. Use when builds fail, Anchor CLI won't install, or CI breaks on Debian\u002FUbuntu.","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fskills\u002Fsolana-common-errors\u002FSKILL.md",[],[],{"skillNo":148,"name":149,"nameZh":6,"description":150,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":151,"sourceHtmlUrl":152,"categories":153,"keywords":154,"publishedAt":19},"01M78B6T7YSYWQ0N4DP40QDXEZ","skill-creator","Create or update AgentSkills. Use when designing, structuring, or packaging skills with scripts, references, and assets.","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fskills\u002Fskill-creator\u002Flicense.txt","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fskills\u002Fskill-creator\u002FSKILL.md",[],[],{"skillNo":156,"name":157,"nameZh":6,"description":158,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":120,"licenseFileUrl":6,"sourceHtmlUrl":159,"categories":160,"keywords":161,"publishedAt":19},"01M78B1VFT9X5AND7AE1HZPZGF","phantom-wallet-mcp","Execute wallet operations through the Phantom MCP server. Use when the user wants to interact with their Phantom wallet directly — get addresses, transfer SOL or SPL tokens, buy\u002Fswap tokens, sign transactions, and sign messages across Solana, Ethereum, Bitcoin, and Sui. Requires the @phantom\u002Fmcp-server to be configured as an MCP server.","https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fskills\u002Fphantom-wallet-mcp\u002FSKILL.md",[],[]]