[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M787PG24TY4FY4STSN7CPPSS":3,"skill-same:01M787PG24TY4FY4STSN7CPPSS":99},{"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},"01M787PG24TY4FY4STSN7CPPSS","earth2studio-create-prognostic",null,"Create Earth2Studio prognostic (time-stepping forecast) model wrappers. Do NOT use for diagnostic models, data sources, or installation.","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\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Fearth2studio-create-prognostic\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: earth2studio-create-prognostic\nversion: 0.16.0\nlicense: Apache-2.0\nmetadata:\n  author: NVIDIA Earth-2 Team \u003Cagent-skills@nvidia.com>\n  tags: [earth2studio, prognostic-model, python]\ndescription: >\n  Create Earth2Studio prognostic (time-stepping forecast) model wrappers.\n  Do NOT use for diagnostic models, data sources, or installation.\nargument-hint: URL or local path to reference inference script (optional)\n---\n\n## Quick Start Checklist\n\n**Do these steps IN ORDER. Do not skip any step.**\n\n- [ ] Read this SKILL.md completely first\n- [ ] Get reference script (Step 0)\n- [ ] Create `earth2studio\u002Fmodels\u002Fpx\u002F\u003Cname>.py` with triple inheritance\n- [ ] Create `test\u002Fmodels\u002Fpx\u002Ftest_\u003Cname>.py` with mock tests\n- [ ] Run: `uv run pytest test\u002Fmodels\u002Fpx\u002Ftest_\u003Cname>.py -v`\n- [ ] Add\u002Fupdate model extra, install docs, API docs, and changelog (Steps 1-2, 9)\n- [ ] Run: `make format && make lint`\n\n> **⚠️ CRITICAL:** Always use `uv run` for Python commands:\n> - ✅ `uv run pytest ...` \u002F `uv run python ...`\n> - ❌ `pytest ...` \u002F `python ...` (missing dependencies)\n>\n> **Stuck or wrong output:** Do not keep retrying the same fix. Follow\n> [Self-Improvement](#self-improvement) to patch this skill before continuing.\n\n## Purpose\n\nImplement a prognostic model wrapper connecting third-party ML weather models\nto Earth2Studio. Prognostic models time-integrate forward—given initial state,\nthey predict future states by stepping through time (e.g., 6-hour increments).\n\n## Workspace\n\n| Context | Location |\n|---------|----------|\n| Harbor eval | Write to `\u002Fworkspace\u002Foutput\u002Fearth2studio\u002Fmodels\u002Fpx\u002F...` |\n| Harbor + `--copy-repo` | Full checkout at `\u002Fworkspace\u002Frepo` |\n| Local clone | Directory with `pyproject.toml` |\n\n**Never read `evals\u002Ftargets\u002F`** — grader references only.\n\n### Reference Files\n\nLoad on demand during the matching step:\n\n| File | Content | Load at |\n|------|---------|---------|\n| `references\u002Fskeleton-template.py` | Full model skeleton with FILL comments | Steps 3–6 |\n| `references\u002Fmethod-templates.py` | Canonical method implementations | Steps 4–6 |\n| `references\u002Ftesting-guide.py` | Test skeleton and mock patterns | Step 7 |\n| `references\u002Fvalidation-guide.md` | Comparison scripts, PR, code review | Steps 10–11 |\n\n---\n\n## Workflow Steps\n\n### Step 0 — Get Reference Script\n\nIf `$ARGUMENTS` provided, use it. Otherwise ask:\n> Please provide a reference inference script URL\u002Fpath.\n\n### Step 1 — Analyze & Propose Dependencies\n\nAnalyze: packages, architecture, I\u002FO shapes, time step, resolution, checkpoint.\n\nPropose `pyproject.toml` group (alphabetical, add to `all`). Every\nprognostic model must have an optional dependency extra, even when no packages\nare required:\n```toml\nmodel-name = [\"package1>=version\", \"package2\"]\n# or, when no additional packages are required:\nmodel-name = []\n```\n\n**[CONFIRM]** Present dependencies and ask user to approve.\n\n### Step 2 — Add Dependencies\n\nEdit `pyproject.toml`: add the model extra alphabetically, even if it is\nempty, and update the `all` aggregate.\n\n### Step 3 — Create Model File\n\n**File:** `earth2studio\u002Fmodels\u002Fpx\u002F\u003Clowercase>.py`\n\n**Required inheritance (all three):**\n```python\nclass ModelName(torch.nn.Module, AutoModelMixin, PrognosticMixin):\n```\n\n**Required imports:**\n```python\nimport numpy as np\nimport torch\nfrom earth2studio.models.auto import AutoModelMixin, Package\nfrom earth2studio.models.batch import batch_coords, batch_func\nfrom earth2studio.models.px.base import PrognosticMixin\nfrom earth2studio.models.utils import create_coords_from_lat_lon, handshake_dim\nfrom earth2studio.lexicon import E2STUDIO_VOCAB\nfrom earth2studio.utils import check_optional_dependencies\nfrom loguru import logger\n```\n\n**SPDX header (required at top of every .py file):**\n```python\n# SPDX-FileCopyrightText: Copyright (c) 2024-2025 NVIDIA CORPORATION & AFFILIATES.\n# SPDX-License-Identifier: Apache-2.0\n```\n\n**Canonical method order:**\n1. `__init__` 2. `input_coords` 3. `output_coords` (@batch_coords)\n4. `load_default_package` 5. `load_model` 6. `to` (optional)\n7. Private methods 8. `__call__` (@batch_func) 9. `_default_generator`\n10. `create_iterator`\n\n### Step 4 — Implement Coordinates\n\n**input_coords rules:**\n- `batch`: `np.empty(0)`\n- `time`: `np.empty(0)` (dynamic)\n- `lead_time`: starts at `np.timedelta64(0, \"h\")`\n- `lat`: 90 to -90 (north to south); this is the public Earth2Studio convention even if the source model uses the opposite order\n- `lon`: 0 to 360\n- If a checkpoint\u002Fmodel core expects south-to-north latitude, flip tensors internally before\u002Fafter the core model; do not expose flipped latitude in `input_coords` or `output_coords`\n- Map variables to `E2STUDIO_VOCAB` (282 entries in `earth2studio\u002Flexicon\u002Fbase.py`)\n\n**output_coords:** Use `handshake_dim`\u002F`handshake_coords` for input validation, then increment `lead_time`. Prefer a shared coordinate-check helper and call it from `output_coords`, `__call__`, and iterator setup before model execution.\n\n### Step 5 — Implement Forward Pass\n\n**`__call__`:** @batch_func decorated, shape (batch, time, lead_time, var, lat, lon).\nReshape to model format → call model → reshape back.\n\n**`create_iterator`:** MUST yield initial condition first (step 0).\nUse `front_hook`\u002F`rear_hook` for perturbation injection.\n\n### Step 6 — Implement Model Loading\n\n**`load_default_package`:** Lock HuggingFace URLs: `hf:\u002F\u002Forg\u002Frepo@commit`\n\n**`load_model`:** Use `package.resolve()`, `map_location=\"cpu\"`, `eval()` mode,\ndecorate with `@check_optional_dependencies()`.\n\n### Step 7 — Write Tests\n\n**File:** `test\u002Fmodels\u002Fpx\u002Ftest_\u003Cname>.py`\n\n**Required tests:**\n| Function | Purpose |\n|----------|---------|\n| `test_\u003Cmodel>_call` | Single forward pass (parametrize device\u002Ftime) |\n| `test_\u003Cmodel>_iter` | Iterator produces sequence |\n| `test_\u003Cmodel>_exceptions` | Invalid coords raise errors |\n| `test_\u003Cmodel>_package` | Real weights (`@pytest.mark.package`) |\n\nCreate `PhooModelName` dummy matching interface for mock tests.\n\n**Run tests:**\n```bash\nuv run pytest test\u002Fmodels\u002Fpx\u002Ftest_\u003Cname>.py -m \"not package\" -v\nuv run pytest test\u002Fmodels\u002Fpx\u002Ftest_\u003Cname>.py::test_\u003Cmodel>_package --package -v\n```\n\nDo not omit the package test. If arbitrary random inputs are not physically\nvalid for the real checkpoint, use a stable model-appropriate synthetic input\nwhile still loading real weights and running a forward pass.\n\n### Step 8 — Register Model (if requested)\n\n- Add to `earth2studio\u002Fmodels\u002Fpx\u002F__init__.py` (alphabetical)\n- Verify deps in pyproject.toml\n\n### Step 9 — Documentation\n\n- Add to `docs\u002Fmodules\u002Fmodels_px.rst` (alphabetical). This is required for\n  every new prognostic model so the API docs include the generated page.\n- Add to `docs\u002Fuserguide\u002Fabout\u002Finstall.md` (alphabetical tab) for the\n  model extra, even when the extra is empty. Include model-specific notes plus\n  both `pip install earth2studio[model-name]` and\n  `uv add earth2studio --extra model-name` instructions.\n- Update `CHANGELOG.md` under `### Added`. This is required for every new\n  prognostic model.\n\n**Format and lint:**\n```bash\nmake format && make lint && make license\n```\n\n### Step 10 - Validation (if requested)\n\nFollow `references\u002Fvalidation-guide.md`. Create uncommitted vanilla, E2S,\ncomparison, and sanity-check scripts; do not commit generated outputs or images.\nUse PR-safe placeholders for plots so the user can upload images manually.\n\n**[CONFIRM]** User must visually inspect plots before proceeding.\n\n### Step 11 - PR (if requested)\n\nFollow `references\u002Fvalidation-guide.md` and use:\n- `references\u002Fpr-body-template.md`\n- `references\u002Fpr-comment-template.md`\n\nBefore creating the PR, verify `pyproject.toml` has the model extra, the\n`all` extra includes it, install docs include both pip and uv commands, and\n`docs\u002Fmodules\u002Fmodels_px.rst` plus `CHANGELOG.md` are updated.\n\nDo not include machine names, absolute paths, device inventory, or uploaded image\nlinks in PR text. Use plot placeholders instead.\n\n---\n\n## Examples\n\n### Simple Identity Model\n```text\nUser: Create IdentityModel - returns input unchanged, 6h step, 181x360, vars: t2m, u10m, v10m, msl\n\nAgent: [reads SKILL.md, creates identity.py with triple inheritance,\n        creates test_identity.py, runs pytest, runs make format && lint]\n```\n\n### External Model (Pangu)\n```text\nUser: Add Pangu-Weather wrapper\n      GitHub: https:\u002F\u002Fgithub.com\u002F198808xc\u002FPangu-Weather\n\nAgent: [reads SKILL.md, fetches inference.py, creates pangu.py,\n        creates test_pangu.py, runs pytest]\n```\n\n---\n\n## Key Patterns\n\n### Coordinate Template\n```python\n@property\ndef input_coords(self) -> CoordSystem:\n    return CoordSystem({\n        \"batch\": np.empty(0),\n        \"time\": np.empty(0),\n        \"lead_time\": np.array([np.timedelta64(0, \"h\")]),\n        \"variable\": np.array([\"t2m\", \"u10m\", ...]),\n        # Public Earth2Studio convention is north-to-south latitude.\n        \"lat\": np.linspace(90, -90, 181),\n        \"lon\": np.linspace(0, 359, 360),\n    })\n\n@batch_coords()\ndef output_coords(self, input_coords: CoordSystem) -> CoordSystem:\n    output = input_coords.copy()\n    output[\"lead_time\"] = input_coords[\"lead_time\"] + np.timedelta64(6, \"h\")\n    return output\n```\n\n### Iterator Template\n```python\ndef create_iterator(self, x, coords):\n    yield x, coords  # Initial condition (step 0)\n    while True:\n        x, coords = self.front_hook(x, coords)\n        x, coords = self(x, coords)\n        x, coords = self.rear_hook(x, coords)\n        yield x, coords\n```\n\n---\n\n## Troubleshooting\n\n| Error | Solution |\n|-------|----------|\n| `OptionalDependencyFailure` | `uv add --optional \u003Cgroup> \u003Cpkg>` |\n| Coordinate handshake fails | Check `handshake_dim` indices match dim position |\n| Iterator wrong shapes | Debug reshape logic with random input |\n| `ModuleNotFoundError: pytest` | Use `uv run pytest` not `pytest` |\n\n---\n\n## Reminders\n\n**DO:**\n- Use `uv run python` for ALL Python commands\n- Use `loguru.logger`, never `print()`\n- Inherit `torch.nn.Module + AutoModelMixin + PrognosticMixin`\n- Yield initial condition first in `create_iterator`\n- Use `front_hook()`\u002F`rear_hook()` in `_default_generator`\n- Include SPDX header in every .py file\n\n**DON'T:**\n- Create general base classes for reuse\n- Commit API keys or comparison scripts\n- Read from `evals\u002Ftargets\u002F`\n","## Quick Start Checklist\n\n**Do these steps IN ORDER. Do not skip any step.**\n\n- [ ] Read this SKILL.md completely first\n- [ ] Get reference script (Step 0)\n- [ ] Create `earth2studio\u002Fmodels\u002Fpx\u002F\u003Cname>.py` with triple inheritance\n- [ ] Create `test\u002Fmodels\u002Fpx\u002Ftest_\u003Cname>.py` with mock tests\n- [ ] Run: `uv run pytest test\u002Fmodels\u002Fpx\u002Ftest_\u003Cname>.py -v`\n- [ ] Add\u002Fupdate model extra, install docs, API docs, and changelog (Steps 1-2, 9)\n- [ ] Run: `make format && make lint`\n\n> **⚠️ CRITICAL:** Always use `uv run` for Python commands:\n> - ✅ `uv run pytest ...` \u002F `uv run python ...`\n> - ❌ `pytest ...` \u002F `python ...` (missing dependencies)\n>\n> **Stuck or wrong output:** Do not keep retrying the same fix. Follow\n> [Self-Improvement](#self-improvement) to patch this skill before continuing.\n\n## Purpose\n\nImplement a prognostic model wrapper connecting third-party ML weather models\nto Earth2Studio. Prognostic models time-integrate forward—given initial state,\nthey predict future states by stepping through time (e.g., 6-hour increments).\n\n## Workspace\n\n| Context | Location |\n|---------|----------|\n| Harbor eval | Write to `\u002Fworkspace\u002Foutput\u002Fearth2studio\u002Fmodels\u002Fpx\u002F...` |\n| Harbor + `--copy-repo` | Full checkout at `\u002Fworkspace\u002Frepo` |\n| Local clone | Directory with `pyproject.toml` |\n\n**Never read `evals\u002Ftargets\u002F`** — grader references only.\n\n### Reference Files\n\nLoad on demand during the matching step:\n\n| File | Content | Load at |\n|------|---------|---------|\n| `references\u002Fskeleton-template.py` | Full model skeleton with FILL comments | Steps 3–6 |\n| `references\u002Fmethod-templates.py` | Canonical method implementations | Steps 4–6 |\n| `references\u002Ftesting-guide.py` | Test skeleton and mock patterns | Step 7 |\n| `references\u002Fvalidation-guide.md` | Comparison scripts, PR, code review | Steps 10–11 |\n\n---\n\n## Workflow Steps\n\n### Step 0 — Get Reference Script\n\nIf `$ARGUMENTS` provided, use it. Otherwise ask:\n> Please provide a reference inference script URL\u002Fpath.\n\n### Step 1 — Analyze & Propose Dependencies\n\nAnalyze: packages, architecture, I\u002FO shapes, time step, resolution, checkpoint.\n\nPropose `pyproject.toml` group (alphabetical, add to `all`). Every\nprognostic model must have an optional dependency extra, even when no packages\nare required:\n```toml\nmodel-name = [\"package1>=version\", \"package2\"]\n# or, when no additional packages are required:\nmodel-name = []\n```\n\n**[CONFIRM]** Present dependencies and ask user to approve.\n\n### Step 2 — Add Dependencies\n\nEdit `pyproject.toml`: add the model extra alphabetically, even if it is\nempty, and update the `all` aggregate.\n\n### Step 3 — Create Model File\n\n**File:** `earth2studio\u002Fmodels\u002Fpx\u002F\u003Clowercase>.py`\n\n**Required inheritance (all three):**\n```python\nclass ModelName(torch.nn.Module, AutoModelMixin, PrognosticMixin):\n```\n\n**Required imports:**\n```python\nimport numpy as np\nimport torch\nfrom earth2studio.models.auto import AutoModelMixin, Package\nfrom earth2studio.models.batch import batch_coords, batch_func\nfrom earth2studio.models.px.base import PrognosticMixin\nfrom earth2studio.models.utils import create_coords_from_lat_lon, handshake_dim\nfrom earth2studio.lexicon import E2STUDIO_VOCAB\nfrom earth2studio.utils import check_optional_dependencies\nfrom loguru import logger\n```\n\n**SPDX header (required at top of every .py file):**\n```python\n# SPDX-FileCopyrightText: Copyright (c) 2024-2025 NVIDIA CORPORATION & AFFILIATES.\n# SPDX-License-Identifier: Apache-2.0\n```\n\n**Canonical method order:**\n1. `__init__` 2. `input_coords` 3. `output_coords` (@batch_coords)\n4. `load_default_package` 5. `load_model` 6. `to` (optional)\n7. Private methods 8. `__call__` (@batch_func) 9. `_default_generator`\n10. `create_iterator`\n\n### Step 4 — Implement Coordinates\n\n**input_coords rules:**\n- `batch`: `np.empty(0)`\n- `time`: `np.empty(0)` (dynamic)\n- `lead_time`: starts at `np.timedelta64(0, \"h\")`\n- `lat`: 90 to -90 (north to south); this is the public Earth2Studio convention even if the source model uses the opposite order\n- `lon`: 0 to 360\n- If a checkpoint\u002Fmodel core expects south-to-north latitude, flip tensors internally before\u002Fafter the core model; do not expose flipped latitude in `input_coords` or `output_coords`\n- Map variables to `E2STUDIO_VOCAB` (282 entries in `earth2studio\u002Flexicon\u002Fbase.py`)\n\n**output_coords:** Use `handshake_dim`\u002F`handshake_coords` for input validation, then increment `lead_time`. Prefer a shared coordinate-check helper and call it from `output_coords`, `__call__`, and iterator setup before model execution.\n\n### Step 5 — Implement Forward Pass\n\n**`__call__`:** @batch_func decorated, shape (batch, time, lead_time, var, lat, lon).\nReshape to model format → call model → reshape back.\n\n**`create_iterator`:** MUST yield initial condition first (step 0).\nUse `front_hook`\u002F`rear_hook` for perturbation injection.\n\n### Step 6 — Implement Model Loading\n\n**`load_default_package`:** Lock HuggingFace URLs: `hf:\u002F\u002Forg\u002Frepo@commit`\n\n**`load_model`:** Use `package.resolve()`, `map_location=\"cpu\"`, `eval()` mode,\ndecorate with `@check_optional_dependencies()`.\n\n### Step 7 — Write Tests\n\n**File:** `test\u002Fmodels\u002Fpx\u002Ftest_\u003Cname>.py`\n\n**Required tests:**\n| Function | Purpose |\n|----------|---------|\n| `test_\u003Cmodel>_call` | Single forward pass (parametrize device\u002Ftime) |\n| `test_\u003Cmodel>_iter` | Iterator produces sequence |\n| `test_\u003Cmodel>_exceptions` | Invalid coords raise errors |\n| `test_\u003Cmodel>_package` | Real weights (`@pytest.mark.package`) |\n\nCreate `PhooModelName` dummy matching interface for mock tests.\n\n**Run tests:**\n```bash\nuv run pytest test\u002Fmodels\u002Fpx\u002Ftest_\u003Cname>.py -m \"not package\" -v\nuv run pytest test\u002Fmodels\u002Fpx\u002Ftest_\u003Cname>.py::test_\u003Cmodel>_package --package -v\n```\n\nDo not omit the package test. If arbitrary random inputs are not physically\nvalid for the real checkpoint, use a stable model-appropriate synthetic input\nwhile still loading real weights and running a forward pass.\n\n### Step 8 — Register Model (if requested)\n\n- Add to `earth2studio\u002Fmodels\u002Fpx\u002F__init__.py` (alphabetical)\n- Verify deps in pyproject.toml\n\n### Step 9 — Documentation\n\n- Add to `docs\u002Fmodules\u002Fmodels_px.rst` (alphabetical). This is required for\n  every new prognostic model so the API docs include the generated page.\n- Add to `docs\u002Fuserguide\u002Fabout\u002Finstall.md` (alphabetical tab) for the\n  model extra, even when the extra is empty. Include model-specific notes plus\n  both `pip install earth2studio[model-name]` and\n  `uv add earth2studio --extra model-name` instructions.\n- Update `CHANGELOG.md` under `### Added`. This is required for every new\n  prognostic model.\n\n**Format and lint:**\n```bash\nmake format && make lint && make license\n```\n\n### Step 10 - Validation (if requested)\n\nFollow `references\u002Fvalidation-guide.md`. Create uncommitted vanilla, E2S,\ncomparison, and sanity-check scripts; do not commit generated outputs or images.\nUse PR-safe placeholders for plots so the user can upload images manually.\n\n**[CONFIRM]** User must visually inspect plots before proceeding.\n\n### Step 11 - PR (if requested)\n\nFollow `references\u002Fvalidation-guide.md` and use:\n- `references\u002Fpr-body-template.md`\n- `references\u002Fpr-comment-template.md`\n\nBefore creating the PR, verify `pyproject.toml` has the model extra, the\n`all` extra includes it, install docs include both pip and uv commands, and\n`docs\u002Fmodules\u002Fmodels_px.rst` plus `CHANGELOG.md` are updated.\n\nDo not include machine names, absolute paths, device inventory, or uploaded image\nlinks in PR text. Use plot placeholders instead.\n\n---\n\n## Examples\n\n### Simple Identity Model\n```text\nUser: Create IdentityModel - returns input unchanged, 6h step, 181x360, vars: t2m, u10m, v10m, msl\n\nAgent: [reads SKILL.md, creates identity.py with triple inheritance,\n        creates test_identity.py, runs pytest, runs make format && lint]\n```\n\n### External Model (Pangu)\n```text\nUser: Add Pangu-Weather wrapper\n      GitHub: https:\u002F\u002Fgithub.com\u002F198808xc\u002FPangu-Weather\n\nAgent: [reads SKILL.md, fetches inference.py, creates pangu.py,\n        creates test_pangu.py, runs pytest]\n```\n\n---\n\n## Key Patterns\n\n### Coordinate Template\n```python\n@property\ndef input_coords(self) -> CoordSystem:\n    return CoordSystem({\n        \"batch\": np.empty(0),\n        \"time\": np.empty(0),\n        \"lead_time\": np.array([np.timedelta64(0, \"h\")]),\n        \"variable\": np.array([\"t2m\", \"u10m\", ...]),\n        # Public Earth2Studio convention is north-to-south latitude.\n        \"lat\": np.linspace(90, -90, 181),\n        \"lon\": np.linspace(0, 359, 360),\n    })\n\n@batch_coords()\ndef output_coords(self, input_coords: CoordSystem) -> CoordSystem:\n    output = input_coords.copy()\n    output[\"lead_time\"] = input_coords[\"lead_time\"] + np.timedelta64(6, \"h\")\n    return output\n```\n\n### Iterator Template\n```python\ndef create_iterator(self, x, coords):\n    yield x, coords  # Initial condition (step 0)\n    while True:\n        x, coords = self.front_hook(x, coords)\n        x, coords = self(x, coords)\n        x, coords = self.rear_hook(x, coords)\n        yield x, coords\n```\n\n---\n\n## Troubleshooting\n\n| Error | Solution |\n|-------|----------|\n| `OptionalDependencyFailure` | `uv add --optional \u003Cgroup> \u003Cpkg>` |\n| Coordinate handshake fails | Check `handshake_dim` indices match dim position |\n| Iterator wrong shapes | Debug reshape logic with random input |\n| `ModuleNotFoundError: pytest` | Use `uv run pytest` not `pytest` |\n\n---\n\n## Reminders\n\n**DO:**\n- Use `uv run python` for ALL Python commands\n- Use `loguru.logger`, never `print()`\n- Inherit `torch.nn.Module + AutoModelMixin + PrognosticMixin`\n- Yield initial condition first in `create_iterator`\n- Use `front_hook()`\u002F`rear_hook()` in `_default_generator`\n- Include SPDX header in every .py file\n\n**DON'T:**\n- Create general base classes for reuse\n- Commit API keys or comparison scripts\n- Read from `evals\u002Ftargets\u002F`\n",{"name":5,"license":15,"version":24,"metadata":25,"description":31,"argument-hint":32},"0.16.0",{"tags":26,"author":30},[27,28,29],"earth2studio","prognostic-model","python","NVIDIA Earth-2 Team \u003Cagent-skills@nvidia.com>","Create Earth2Studio prognostic (time-stepping forecast) model wrappers. Do NOT use for diagnostic models, data sources, or installation.\n","URL or local path to reference inference script (optional)","main","86cc27ffd16af6cb85948f1155578e6f3f61cc4c","skillhub\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Fearth2studio-create-prognostic","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 earth2studio-create-prognostic",{"tool":47,"command":48},"bunx","bunx skills add Solizardking\u002Ftrain2earn --skill earth2studio-create-prognostic",{"tool":50,"command":51},"pnpm","pnpm dlx skills add Solizardking\u002Ftrain2earn --skill earth2studio-create-prognostic","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Fearth2studio-create-prognostic\u002FSKILL.md",true,[55,61,65,71,75,79,83,87,91,95],{"path":56,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":59,"sourceHtmlUrl":60},"BENCHMARK.md","REFERENCE","参考文档",4385,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Fearth2studio-create-prognostic\u002FBENCHMARK.md",{"path":62,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":63,"sourceHtmlUrl":64},"metadata.json",376,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Fearth2studio-create-prognostic\u002Fmetadata.json",{"path":66,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":69,"sourceHtmlUrl":70},"references\u002Fmethod-templates.py","SCRIPT","脚本",12915,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Fearth2studio-create-prognostic\u002Freferences\u002Fmethod-templates.py",{"path":72,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":73,"sourceHtmlUrl":74},"references\u002Fpr-body-template.md",2496,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Fearth2studio-create-prognostic\u002Freferences\u002Fpr-body-template.md",{"path":76,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":77,"sourceHtmlUrl":78},"references\u002Fpr-comment-template.md",2217,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Fearth2studio-create-prognostic\u002Freferences\u002Fpr-comment-template.md",{"path":80,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":81,"sourceHtmlUrl":82},"references\u002Fskeleton-template.py",15570,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Fearth2studio-create-prognostic\u002Freferences\u002Fskeleton-template.py",{"path":84,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":85,"sourceHtmlUrl":86},"references\u002Ftesting-guide.py",14061,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Fearth2studio-create-prognostic\u002Freferences\u002Ftesting-guide.py",{"path":88,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":89,"sourceHtmlUrl":90},"references\u002Fvalidation-guide.md",4935,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Fearth2studio-create-prognostic\u002Freferences\u002Fvalidation-guide.md",{"path":92,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":93,"sourceHtmlUrl":94},"skill-card.md",2493,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Fearth2studio-create-prognostic\u002Fskill-card.md",{"path":96,"resourceType":57,"resourceTypeDesc":58,"sizeBytes":97,"sourceHtmlUrl":98},"verification.json",2411,"https:\u002F\u002Fgithub.com\u002FSolizardking\u002Ftrain2earn\u002Fblob\u002Fmain\u002Fskillhub\u002Fpublic\u002Fapi\u002Fskills\u002Fnvidia\u002Fearth2studio-create-prognostic\u002Fverification.json",[100,108,116,124,131,139],{"skillNo":101,"name":102,"nameZh":6,"description":103,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":104,"licenseFileUrl":6,"sourceHtmlUrl":105,"categories":106,"keywords":107,"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":109,"name":110,"nameZh":6,"description":111,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":104,"licenseFileUrl":112,"sourceHtmlUrl":113,"categories":114,"keywords":115,"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":117,"name":118,"nameZh":6,"description":119,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":104,"licenseFileUrl":120,"sourceHtmlUrl":121,"categories":122,"keywords":123,"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":125,"name":126,"nameZh":6,"description":127,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":104,"licenseFileUrl":6,"sourceHtmlUrl":128,"categories":129,"keywords":130,"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":132,"name":133,"nameZh":6,"description":134,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":135,"sourceHtmlUrl":136,"categories":137,"keywords":138,"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":140,"name":141,"nameZh":6,"description":142,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":104,"licenseFileUrl":6,"sourceHtmlUrl":143,"categories":144,"keywords":145,"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",[],[]]