[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-microsoft-msresearch-aurora":3,"mdc--oysj8g-key":34,"related-repo-microsoft-msresearch-aurora":1113,"related-org-microsoft-msresearch-aurora":1153},{"slug":4,"name":4,"fn":5,"description":6,"org":7,"tags":11,"stars":23,"repoUrl":24,"updatedAt":25,"license":26,"forks":27,"topics":28,"repo":29,"sourceUrl":32,"mdContent":33},"msresearch-aurora","forecast weather with Aurora foundation model","Aurora — Microsoft's foundation model for the Earth system. Use this skill for both inference (regional weather\u002Fclimate forecasting, Norway example, ERA5\u002FCDS data, wind\u002Ftemperature\u002Fpressure mapping) and fine-tuning Aurora on new variables, regions, or domains (training, evaluation, checkpoint extension, gradient\u002Floss debugging). Triggers on Aurora, Microsoft Aurora, foundation models for Earth systems, ERA5\u002FCDS\u002FNetCDF weather data, weather forecast validation, and adapting Aurora to new use cases.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},"microsoft","Microsoft","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Fmicrosoft.png",[12,16,19,20],{"name":13,"slug":14,"type":15},"Research","research","tag",{"name":17,"slug":18,"type":15},"LLM","llm",{"name":9,"slug":8,"type":15},{"name":21,"slug":22,"type":15},"Physics","physics",34,"https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Fvibe-kit","2026-07-03T16:32:28.677219","MIT",7,[],{"repoUrl":24,"stars":23,"forks":27,"topics":30,"description":31},[],"Microsoft Research Vibe Kit is a customized vibe coding environment to enable faster prototyping and integration of Microsoft Research innovations using GitHub Copilot inside Visual Studio Code. ","https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Fvibe-kit\u002Ftree\u002FHEAD\u002Fskills\u002Fmsresearch-aurora","---\nname: msresearch-aurora\ndescription: Aurora — Microsoft's foundation model for the Earth system. Use this skill for both inference (regional weather\u002Fclimate forecasting, Norway example, ERA5\u002FCDS data, wind\u002Ftemperature\u002Fpressure mapping) and fine-tuning Aurora on new variables, regions, or domains (training, evaluation, checkpoint extension, gradient\u002Floss debugging). Triggers on Aurora, Microsoft Aurora, foundation models for Earth systems, ERA5\u002FCDS\u002FNetCDF weather data, weather forecast validation, and adapting Aurora to new use cases.\nlicense: MIT\n---\n## Choosing the right path\n\nBefore diving in, identify what the user is actually trying to do. Three buckets:\n\n1. **\"I want to actually use Aurora for my problem\"** (custom variables, regional adaptation, domain-specific predictions) → **fine-tuning path**. This is the recommended path for serious applications.\n   - Route: [docs\u002Fabout-finetune.md](docs\u002Fabout-finetune.md) → [docs\u002Fquick-start-finetune.md](docs\u002Fquick-start-finetune.md) → [docs\u002Ffinetuning-guide.md](docs\u002Ffinetuning-guide.md).\n\n2. **\"I'm exploring \u002F just curious \u002F want to see Aurora work\"** → **inference path** (Norway regional example with frontend visualization). Lower commitment, no GPU required.\n   - Route: [docs\u002Fabout-aurora.md](docs\u002Fabout-aurora.md) → [docs\u002Fquick-start-inference.md](docs\u002Fquick-start-inference.md) → [docs\u002Fcustomization-guide.md](docs\u002Fcustomization-guide.md) for adapting to a new region.\n\n3. **\"Not sure \u002F want to see Aurora in action first\"** → route by audience:\n   - **Non-technical audience** (execs, PMs, general scientists): start with the **Norway Prototype** ([docs\u002Fquick-start-inference.md](docs\u002Fquick-start-inference.md)) — visual maps over a real region, intuitive at a glance.\n   - **Weather\u002Fclimate practitioners** (familiar with NWP, ERA5, model evaluation): start with the **Finetune Exploration Demo** ([docs\u002Ffinetune-demo.md](docs\u002Ffinetune-demo.md)) — interactive divergence metrics and batch comparisons that surface fine-tuning behavior.\n   Once they have a use case in mind, steer them toward the fine-tuning path.\n\nIf a user mentions a real problem they want Aurora to solve (custom forecasting, air quality, wave heights, regional energy planning, etc.), check the user for their familiarity\u002Fcomfort with the Aurora model and finetuning in general. If a user is comfortable, steer them toward fine-tuning Aurora. If they want to learn more about Aurora first, direct them to [docs\u002Fabout-aurora.md](docs\u002Fabout-aurora.md). If they're familiar with Aurora and\u002For weather modeling, but they're new to finetuning, send them to [docs\u002Fabout-finetune.md](docs\u002Fabout-finetune.md).\n\n## Scope\n\n- Local and Azure-assisted Aurora inference for regional forecasts (Norway example as canonical starting point).\n- End-to-end fine-tuning of the base Aurora checkpoint on custom variables, regions, or lead times.\n- Data preparation (ERA5\u002FCDS download, validation, batch construction).\n- Evaluation, checkpoint selection, training instability debugging.\n\n## Prerequisites\n\n- **Windows users:** run inside a **WSL2** distro (Ubuntu recommended) with NVIDIA drivers installed on the Windows host (not inside WSL). The Aurora prototypes and demoes were developed and tested on WSL2\u002FUbuntu; native Windows Python\u002FPowerShell is not supported. See [Microsoft's WSL install guide](https:\u002F\u002Flearn.microsoft.com\u002Fwindows\u002Fwsl\u002Finstall) and [NVIDIA's CUDA-on-WSL guide](https:\u002F\u002Fdocs.nvidia.com\u002Fcuda\u002Fwsl-user-guide\u002Findex.html).\n- Python 3.11+ with [`uv`](https:\u002F\u002Fdocs.astral.sh\u002Fuv\u002F) package manager\n- `torch`, `lightning`, `numpy`, `xarray`, `microsoft-aurora>=1.5.2`\n- Node.js 18+ with `npm` (only for the inference Norway example frontend)\n- 6 GB free disk space (Aurora checkpoint ~5 GB)\n- GPU optional for inference smoke runs; A100-class strongly recommended for serious training. See [docs\u002Fabout-finetune.md](docs\u002Fabout-finetune.md#compute-requirements) for the full hardware matrix.\n\n## Workflow\n\n1. **Decide the path** (see \"Choosing the right path\" above) before touching files.\n2. **Inference path:**\n   1. Load [docs\u002Fabout-aurora.md](docs\u002Fabout-aurora.md) for background if needed.\n   2. Walk through [docs\u002Fquick-start-inference.md](docs\u002Fquick-start-inference.md) to run the Norway example end-to-end.\n   3. Use [docs\u002Fcustomization-guide.md](docs\u002Fcustomization-guide.md) for regional adaptation; [docs\u002Ftechnical-reference.md](docs\u002Ftechnical-reference.md) for grid\u002Fdata\u002Fdeployment specs.\n3. **Fine-tuning path:**\n   1. Run `initialization\u002Finitialize_starter_code.py` first (use `--skip-tests` by default). This copies `starter-code\u002F` into `aurora-finetune\u002F` at the repo root.\n   2. Validate with the smoke test from `aurora-finetune\u002F`: `uv run pytest tests\u002Ftest_training.py::test_finetuning_2t_var_pretrained --maxfail=1`.\n   3. Walk the user through [docs\u002Fquick-start-finetune.md](docs\u002Fquick-start-finetune.md), then deeper material in [docs\u002Ffinetuning-guide.md](docs\u002Ffinetuning-guide.md), [docs\u002Fform-of-a-batch.md](docs\u002Fform-of-a-batch.md), and [docs\u002Favailable-models.md](docs\u002Favailable-models.md).\n4. **Demo path (exploratory):** Route to [docs\u002Ffinetune-demo.md](docs\u002Ffinetune-demo.md) and offer to walk the user through `docker compose up`.\n5. Route all errors to [docs\u002Ftroubleshooting.md](docs\u002Ftroubleshooting.md) — it has dedicated inference and fine-tuning sections.\n\n### Operational notes\n\n- **Terminal hygiene (inference):** During the Norway quick-start, the frontend dev server (Step 1) must keep running in its own terminal. Open a *new* terminal for Steps 2-5 (Python install, inference, build) so you don't kill the dev server.\n- **Repo-root trap (inference):** `npm install` and `npm run dev` must run from `assets\u002Finference\u002Fnorway-example\u002Ffrontend\u002F` (or the analogous regional directory). The repo root has a different boilerplate app and will fail with confusing errors.\n- **Install `cdsapi` proactively** as soon as the user mentions adapting to a new region or downloading ERA5 data: `uv pip install cdsapi`. Don't wait for an import error mid-workflow.\n- **Secret hygiene:** When walking the user through CDS credential setup (`assets\u002Finference\u002F.env.example` → `.env`), instruct them to edit `.env` locally. Never accept, display, or commit API keys.\n- **Execute, don't display:** When the agent supports terminal execution, run quick-start commands directly via the terminal tool rather than displaying bash blocks for the user to copy. Show bash blocks only when reviewing or when the user explicitly wants to copy commands.\n- **Fine-tuning quick-start UX:** Brief the user before each command, confirm before continuing, and summarize completed vs. next actions.\n- **When a user starts or completes fine-tuning, surface TensorBoard:** Remind them to inspect loss curves with `uv run tensorboard --logdir tb_logs\u002Ffinetuning` and open `http:\u002F\u002Flocalhost:6006`. Look for steady descent in `train_loss` and `val_loss`; flag plateauing or divergence with concrete remediation prompts (see `docs\u002Ftroubleshooting.md`).\n- **After any code change in the fine-tuning workflow:** Run `uv run pytest` (or a targeted eval) before declaring success. Flag NaN, divergence, or plateauing loss with concrete remediation prompts (lower learning rate, gradient clipping, see `docs\u002Ftroubleshooting.md`).\n\n## Routing\n\nFormat: `{{Doc}}` — {{when to load}}\n\n- [docs\u002Fabout-aurora.md](docs\u002Fabout-aurora.md) — User asks what Aurora is, how it works, why it matters, or needs background.\n- [docs\u002Fabout-finetune.md](docs\u002Fabout-finetune.md) — User asks why fine-tune Aurora, who fine-tuning is for, or what it unlocks.\n- [docs\u002Ffinetune-demo.md](docs\u002Ffinetune-demo.md) — User wants to see Aurora outputs interactively without training (best for weather\u002Fclimate practitioners).\n- [docs\u002Fquick-start-inference.md](docs\u002Fquick-start-inference.md) — User wants to run the Norway forecast or visualize Aurora predictions (best for non-technical audiences and first-look demos).\n- [docs\u002Fquick-start-finetune.md](docs\u002Fquick-start-finetune.md) — User wants to run a first fine-tuning experiment on bundled data.\n- [docs\u002Fcustomization-guide.md](docs\u002Fcustomization-guide.md) — User wants a new region for the inference example, custom prototype, or domain recipes.\n- [docs\u002Ffinetuning-guide.md](docs\u002Ffinetuning-guide.md) — Training\u002Foptimization, gradients, AMP, LoRA, variable extension.\n- [docs\u002Fform-of-a-batch.md](docs\u002Fform-of-a-batch.md) — Data format \u002F `aurora.Batch` structure.\n- [docs\u002Favailable-models.md](docs\u002Favailable-models.md) — Checkpoint selection.\n- [docs\u002Finference-basics.md](docs\u002Finference-basics.md) — Running predictions without the Norway frontend.\n- [docs\u002Ftechnical-reference.md](docs\u002Ftechnical-reference.md) — Grid constraints, data formats, variable naming, deployment, validation.\n- [docs\u002Ftroubleshooting.md](docs\u002Ftroubleshooting.md) — Errors and diagnostics for both inference and fine-tuning.\n- [starter-code\u002FREADME.md](starter-code\u002FREADME.md) — CLI and starter-code reference.\n- [assets\u002Finference\u002Fscripts\u002FREADME.md](assets\u002Finference\u002Fscripts\u002FREADME.md) — Inference scripts and utilities.\n\n### Assets\n\n- `assets\u002Finference\u002Fnorway-example\u002F` — Reference inference implementation with frontend.\n- `assets\u002Finference\u002Fscripts\u002F` — ERA5 download, validation, regional setup utilities.\n- `assets\u002Finference\u002Fpaper\u002F` — Paper artifacts and supplementary material.\n- `assets\u002Ffinetune\u002Ffinetune-exploration-demo-app\u002F` — Dockerized exploration demo (no training required).\n- `starter-code\u002F` — Full Python project (`vibe-tune-aurora`) with training loop, losses, data loaders, CLIs, and tests.\n- `initialization\u002Finitialize_starter_code.py` — Copies `starter-code\u002F` into the user's repo as `aurora-finetune\u002F`.\n\n## Reference Links\n\n- Research Paper: [A foundation model for the Earth system](https:\u002F\u002Fwww.nature.com\u002Farticles\u002Fs41586-025-09005-y) (Nature, 2025)\n- GitHub: [github.com\u002Fmicrosoft\u002Faurora](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Faurora)\n- Model Weights: [huggingface.co\u002Fmicrosoft\u002Faurora](https:\u002F\u002Fhuggingface.co\u002Fmicrosoft\u002Faurora)\n- Azure AI Foundry: [ai.azure.com\u002Fcatalog\u002Fmodels\u002FAurora](https:\u002F\u002Fai.azure.com\u002Fcatalog\u002Fmodels\u002FAurora)\n",{"data":35,"body":36},{"name":4,"description":6,"license":26},{"type":37,"children":38},"root",[39,48,54,208,224,230,253,259,384,390,585,592,796,802,820,960,966,1054,1060],{"type":40,"tag":41,"props":42,"children":44},"element","h2",{"id":43},"choosing-the-right-path",[45],{"type":46,"value":47},"text","Choosing the right path",{"type":40,"tag":49,"props":50,"children":51},"p",{},[52],{"type":46,"value":53},"Before diving in, identify what the user is actually trying to do. Three buckets:",{"type":40,"tag":55,"props":56,"children":57},"ol",{},[58,107,149],{"type":40,"tag":59,"props":60,"children":61},"li",{},[62,68,70,75,77],{"type":40,"tag":63,"props":64,"children":65},"strong",{},[66],{"type":46,"value":67},"\"I want to actually use Aurora for my problem\"",{"type":46,"value":69}," (custom variables, regional adaptation, domain-specific predictions) → ",{"type":40,"tag":63,"props":71,"children":72},{},[73],{"type":46,"value":74},"fine-tuning path",{"type":46,"value":76},". This is the recommended path for serious applications.",{"type":40,"tag":78,"props":79,"children":80},"ul",{},[81],{"type":40,"tag":59,"props":82,"children":83},{},[84,86,92,94,99,100,105],{"type":46,"value":85},"Route: ",{"type":40,"tag":87,"props":88,"children":90},"a",{"href":89},"docs\u002Fabout-finetune.md",[91],{"type":46,"value":89},{"type":46,"value":93}," → ",{"type":40,"tag":87,"props":95,"children":97},{"href":96},"docs\u002Fquick-start-finetune.md",[98],{"type":46,"value":96},{"type":46,"value":93},{"type":40,"tag":87,"props":101,"children":103},{"href":102},"docs\u002Ffinetuning-guide.md",[104],{"type":46,"value":102},{"type":46,"value":106},".",{"type":40,"tag":59,"props":108,"children":109},{},[110,115,116,121,123],{"type":40,"tag":63,"props":111,"children":112},{},[113],{"type":46,"value":114},"\"I'm exploring \u002F just curious \u002F want to see Aurora work\"",{"type":46,"value":93},{"type":40,"tag":63,"props":117,"children":118},{},[119],{"type":46,"value":120},"inference path",{"type":46,"value":122}," (Norway regional example with frontend visualization). Lower commitment, no GPU required.",{"type":40,"tag":78,"props":124,"children":125},{},[126],{"type":40,"tag":59,"props":127,"children":128},{},[129,130,135,136,141,142,147],{"type":46,"value":85},{"type":40,"tag":87,"props":131,"children":133},{"href":132},"docs\u002Fabout-aurora.md",[134],{"type":46,"value":132},{"type":46,"value":93},{"type":40,"tag":87,"props":137,"children":139},{"href":138},"docs\u002Fquick-start-inference.md",[140],{"type":46,"value":138},{"type":46,"value":93},{"type":40,"tag":87,"props":143,"children":145},{"href":144},"docs\u002Fcustomization-guide.md",[146],{"type":46,"value":144},{"type":46,"value":148}," for adapting to a new region.",{"type":40,"tag":59,"props":150,"children":151},{},[152,157,159],{"type":40,"tag":63,"props":153,"children":154},{},[155],{"type":46,"value":156},"\"Not sure \u002F want to see Aurora in action first\"",{"type":46,"value":158}," → route by audience:",{"type":40,"tag":78,"props":160,"children":161},{},[162,185],{"type":40,"tag":59,"props":163,"children":164},{},[165,170,172,177,179,183],{"type":40,"tag":63,"props":166,"children":167},{},[168],{"type":46,"value":169},"Non-technical audience",{"type":46,"value":171}," (execs, PMs, general scientists): start with the ",{"type":40,"tag":63,"props":173,"children":174},{},[175],{"type":46,"value":176},"Norway Prototype",{"type":46,"value":178}," (",{"type":40,"tag":87,"props":180,"children":181},{"href":138},[182],{"type":46,"value":138},{"type":46,"value":184},") — visual maps over a real region, intuitive at a glance.",{"type":40,"tag":59,"props":186,"children":187},{},[188,193,195,200,201,206],{"type":40,"tag":63,"props":189,"children":190},{},[191],{"type":46,"value":192},"Weather\u002Fclimate practitioners",{"type":46,"value":194}," (familiar with NWP, ERA5, model evaluation): start with the ",{"type":40,"tag":63,"props":196,"children":197},{},[198],{"type":46,"value":199},"Finetune Exploration Demo",{"type":46,"value":178},{"type":40,"tag":87,"props":202,"children":204},{"href":203},"docs\u002Ffinetune-demo.md",[205],{"type":46,"value":203},{"type":46,"value":207},") — interactive divergence metrics and batch comparisons that surface fine-tuning behavior.\nOnce they have a use case in mind, steer them toward the fine-tuning path.",{"type":40,"tag":49,"props":209,"children":210},{},[211,213,217,219,223],{"type":46,"value":212},"If a user mentions a real problem they want Aurora to solve (custom forecasting, air quality, wave heights, regional energy planning, etc.), check the user for their familiarity\u002Fcomfort with the Aurora model and finetuning in general. If a user is comfortable, steer them toward fine-tuning Aurora. If they want to learn more about Aurora first, direct them to ",{"type":40,"tag":87,"props":214,"children":215},{"href":132},[216],{"type":46,"value":132},{"type":46,"value":218},". If they're familiar with Aurora and\u002For weather modeling, but they're new to finetuning, send them to ",{"type":40,"tag":87,"props":220,"children":221},{"href":89},[222],{"type":46,"value":89},{"type":46,"value":106},{"type":40,"tag":41,"props":225,"children":227},{"id":226},"scope",[228],{"type":46,"value":229},"Scope",{"type":40,"tag":78,"props":231,"children":232},{},[233,238,243,248],{"type":40,"tag":59,"props":234,"children":235},{},[236],{"type":46,"value":237},"Local and Azure-assisted Aurora inference for regional forecasts (Norway example as canonical starting point).",{"type":40,"tag":59,"props":239,"children":240},{},[241],{"type":46,"value":242},"End-to-end fine-tuning of the base Aurora checkpoint on custom variables, regions, or lead times.",{"type":40,"tag":59,"props":244,"children":245},{},[246],{"type":46,"value":247},"Data preparation (ERA5\u002FCDS download, validation, batch construction).",{"type":40,"tag":59,"props":249,"children":250},{},[251],{"type":46,"value":252},"Evaluation, checkpoint selection, training instability debugging.",{"type":40,"tag":41,"props":254,"children":256},{"id":255},"prerequisites",[257],{"type":46,"value":258},"Prerequisites",{"type":40,"tag":78,"props":260,"children":261},{},[262,297,316,354,367,372],{"type":40,"tag":59,"props":263,"children":264},{},[265,270,272,277,279,287,289,296],{"type":40,"tag":63,"props":266,"children":267},{},[268],{"type":46,"value":269},"Windows users:",{"type":46,"value":271}," run inside a ",{"type":40,"tag":63,"props":273,"children":274},{},[275],{"type":46,"value":276},"WSL2",{"type":46,"value":278}," distro (Ubuntu recommended) with NVIDIA drivers installed on the Windows host (not inside WSL). The Aurora prototypes and demoes were developed and tested on WSL2\u002FUbuntu; native Windows Python\u002FPowerShell is not supported. See ",{"type":40,"tag":87,"props":280,"children":284},{"href":281,"rel":282},"https:\u002F\u002Flearn.microsoft.com\u002Fwindows\u002Fwsl\u002Finstall",[283],"nofollow",[285],{"type":46,"value":286},"Microsoft's WSL install guide",{"type":46,"value":288}," and ",{"type":40,"tag":87,"props":290,"children":293},{"href":291,"rel":292},"https:\u002F\u002Fdocs.nvidia.com\u002Fcuda\u002Fwsl-user-guide\u002Findex.html",[283],[294],{"type":46,"value":295},"NVIDIA's CUDA-on-WSL guide",{"type":46,"value":106},{"type":40,"tag":59,"props":298,"children":299},{},[300,302,314],{"type":46,"value":301},"Python 3.11+ with ",{"type":40,"tag":87,"props":303,"children":306},{"href":304,"rel":305},"https:\u002F\u002Fdocs.astral.sh\u002Fuv\u002F",[283],[307],{"type":40,"tag":308,"props":309,"children":311},"code",{"className":310},[],[312],{"type":46,"value":313},"uv",{"type":46,"value":315}," package manager",{"type":40,"tag":59,"props":317,"children":318},{},[319,325,327,333,334,340,341,347,348],{"type":40,"tag":308,"props":320,"children":322},{"className":321},[],[323],{"type":46,"value":324},"torch",{"type":46,"value":326},", ",{"type":40,"tag":308,"props":328,"children":330},{"className":329},[],[331],{"type":46,"value":332},"lightning",{"type":46,"value":326},{"type":40,"tag":308,"props":335,"children":337},{"className":336},[],[338],{"type":46,"value":339},"numpy",{"type":46,"value":326},{"type":40,"tag":308,"props":342,"children":344},{"className":343},[],[345],{"type":46,"value":346},"xarray",{"type":46,"value":326},{"type":40,"tag":308,"props":349,"children":351},{"className":350},[],[352],{"type":46,"value":353},"microsoft-aurora>=1.5.2",{"type":40,"tag":59,"props":355,"children":356},{},[357,359,365],{"type":46,"value":358},"Node.js 18+ with ",{"type":40,"tag":308,"props":360,"children":362},{"className":361},[],[363],{"type":46,"value":364},"npm",{"type":46,"value":366}," (only for the inference Norway example frontend)",{"type":40,"tag":59,"props":368,"children":369},{},[370],{"type":46,"value":371},"6 GB free disk space (Aurora checkpoint ~5 GB)",{"type":40,"tag":59,"props":373,"children":374},{},[375,377,382],{"type":46,"value":376},"GPU optional for inference smoke runs; A100-class strongly recommended for serious training. See ",{"type":40,"tag":87,"props":378,"children":380},{"href":379},"docs\u002Fabout-finetune.md#compute-requirements",[381],{"type":46,"value":89},{"type":46,"value":383}," for the full hardware matrix.",{"type":40,"tag":41,"props":385,"children":387},{"id":386},"workflow",[388],{"type":46,"value":389},"Workflow",{"type":40,"tag":55,"props":391,"children":392},{},[393,403,454,550,573],{"type":40,"tag":59,"props":394,"children":395},{},[396,401],{"type":40,"tag":63,"props":397,"children":398},{},[399],{"type":46,"value":400},"Decide the path",{"type":46,"value":402}," (see \"Choosing the right path\" above) before touching files.",{"type":40,"tag":59,"props":404,"children":405},{},[406,411],{"type":40,"tag":63,"props":407,"children":408},{},[409],{"type":46,"value":410},"Inference path:",{"type":40,"tag":55,"props":412,"children":413},{},[414,425,436],{"type":40,"tag":59,"props":415,"children":416},{},[417,419,423],{"type":46,"value":418},"Load ",{"type":40,"tag":87,"props":420,"children":421},{"href":132},[422],{"type":46,"value":132},{"type":46,"value":424}," for background if needed.",{"type":40,"tag":59,"props":426,"children":427},{},[428,430,434],{"type":46,"value":429},"Walk through ",{"type":40,"tag":87,"props":431,"children":432},{"href":138},[433],{"type":46,"value":138},{"type":46,"value":435}," to run the Norway example end-to-end.",{"type":40,"tag":59,"props":437,"children":438},{},[439,441,445,447,452],{"type":46,"value":440},"Use ",{"type":40,"tag":87,"props":442,"children":443},{"href":144},[444],{"type":46,"value":144},{"type":46,"value":446}," for regional adaptation; ",{"type":40,"tag":87,"props":448,"children":450},{"href":449},"docs\u002Ftechnical-reference.md",[451],{"type":46,"value":449},{"type":46,"value":453}," for grid\u002Fdata\u002Fdeployment specs.",{"type":40,"tag":59,"props":455,"children":456},{},[457,462],{"type":40,"tag":63,"props":458,"children":459},{},[460],{"type":46,"value":461},"Fine-tuning path:",{"type":40,"tag":55,"props":463,"children":464},{},[465,502,521],{"type":40,"tag":59,"props":466,"children":467},{},[468,470,476,478,484,486,492,494,500],{"type":46,"value":469},"Run ",{"type":40,"tag":308,"props":471,"children":473},{"className":472},[],[474],{"type":46,"value":475},"initialization\u002Finitialize_starter_code.py",{"type":46,"value":477}," first (use ",{"type":40,"tag":308,"props":479,"children":481},{"className":480},[],[482],{"type":46,"value":483},"--skip-tests",{"type":46,"value":485}," by default). This copies ",{"type":40,"tag":308,"props":487,"children":489},{"className":488},[],[490],{"type":46,"value":491},"starter-code\u002F",{"type":46,"value":493}," into ",{"type":40,"tag":308,"props":495,"children":497},{"className":496},[],[498],{"type":46,"value":499},"aurora-finetune\u002F",{"type":46,"value":501}," at the repo root.",{"type":40,"tag":59,"props":503,"children":504},{},[505,507,512,514,520],{"type":46,"value":506},"Validate with the smoke test from ",{"type":40,"tag":308,"props":508,"children":510},{"className":509},[],[511],{"type":46,"value":499},{"type":46,"value":513},": ",{"type":40,"tag":308,"props":515,"children":517},{"className":516},[],[518],{"type":46,"value":519},"uv run pytest tests\u002Ftest_training.py::test_finetuning_2t_var_pretrained --maxfail=1",{"type":46,"value":106},{"type":40,"tag":59,"props":522,"children":523},{},[524,526,530,532,536,537,542,544,549],{"type":46,"value":525},"Walk the user through ",{"type":40,"tag":87,"props":527,"children":528},{"href":96},[529],{"type":46,"value":96},{"type":46,"value":531},", then deeper material in ",{"type":40,"tag":87,"props":533,"children":534},{"href":102},[535],{"type":46,"value":102},{"type":46,"value":326},{"type":40,"tag":87,"props":538,"children":540},{"href":539},"docs\u002Fform-of-a-batch.md",[541],{"type":46,"value":539},{"type":46,"value":543},", and ",{"type":40,"tag":87,"props":545,"children":547},{"href":546},"docs\u002Favailable-models.md",[548],{"type":46,"value":546},{"type":46,"value":106},{"type":40,"tag":59,"props":551,"children":552},{},[553,558,560,564,566,572],{"type":40,"tag":63,"props":554,"children":555},{},[556],{"type":46,"value":557},"Demo path (exploratory):",{"type":46,"value":559}," Route to ",{"type":40,"tag":87,"props":561,"children":562},{"href":203},[563],{"type":46,"value":203},{"type":46,"value":565}," and offer to walk the user through ",{"type":40,"tag":308,"props":567,"children":569},{"className":568},[],[570],{"type":46,"value":571},"docker compose up",{"type":46,"value":106},{"type":40,"tag":59,"props":574,"children":575},{},[576,578,583],{"type":46,"value":577},"Route all errors to ",{"type":40,"tag":87,"props":579,"children":581},{"href":580},"docs\u002Ftroubleshooting.md",[582],{"type":46,"value":580},{"type":46,"value":584}," — it has dedicated inference and fine-tuning sections.",{"type":40,"tag":586,"props":587,"children":589},"h3",{"id":588},"operational-notes",[590],{"type":46,"value":591},"Operational notes",{"type":40,"tag":78,"props":593,"children":594},{},[595,613,646,672,704,714,724,772],{"type":40,"tag":59,"props":596,"children":597},{},[598,603,605,611],{"type":40,"tag":63,"props":599,"children":600},{},[601],{"type":46,"value":602},"Terminal hygiene (inference):",{"type":46,"value":604}," During the Norway quick-start, the frontend dev server (Step 1) must keep running in its own terminal. Open a ",{"type":40,"tag":606,"props":607,"children":608},"em",{},[609],{"type":46,"value":610},"new",{"type":46,"value":612}," terminal for Steps 2-5 (Python install, inference, build) so you don't kill the dev server.",{"type":40,"tag":59,"props":614,"children":615},{},[616,621,623,629,630,636,638,644],{"type":40,"tag":63,"props":617,"children":618},{},[619],{"type":46,"value":620},"Repo-root trap (inference):",{"type":46,"value":622}," ",{"type":40,"tag":308,"props":624,"children":626},{"className":625},[],[627],{"type":46,"value":628},"npm install",{"type":46,"value":288},{"type":40,"tag":308,"props":631,"children":633},{"className":632},[],[634],{"type":46,"value":635},"npm run dev",{"type":46,"value":637}," must run from ",{"type":40,"tag":308,"props":639,"children":641},{"className":640},[],[642],{"type":46,"value":643},"assets\u002Finference\u002Fnorway-example\u002Ffrontend\u002F",{"type":46,"value":645}," (or the analogous regional directory). The repo root has a different boilerplate app and will fail with confusing errors.",{"type":40,"tag":59,"props":647,"children":648},{},[649,662,664,670],{"type":40,"tag":63,"props":650,"children":651},{},[652,654,660],{"type":46,"value":653},"Install ",{"type":40,"tag":308,"props":655,"children":657},{"className":656},[],[658],{"type":46,"value":659},"cdsapi",{"type":46,"value":661}," proactively",{"type":46,"value":663}," as soon as the user mentions adapting to a new region or downloading ERA5 data: ",{"type":40,"tag":308,"props":665,"children":667},{"className":666},[],[668],{"type":46,"value":669},"uv pip install cdsapi",{"type":46,"value":671},". Don't wait for an import error mid-workflow.",{"type":40,"tag":59,"props":673,"children":674},{},[675,680,682,688,689,695,697,702],{"type":40,"tag":63,"props":676,"children":677},{},[678],{"type":46,"value":679},"Secret hygiene:",{"type":46,"value":681}," When walking the user through CDS credential setup (",{"type":40,"tag":308,"props":683,"children":685},{"className":684},[],[686],{"type":46,"value":687},"assets\u002Finference\u002F.env.example",{"type":46,"value":93},{"type":40,"tag":308,"props":690,"children":692},{"className":691},[],[693],{"type":46,"value":694},".env",{"type":46,"value":696},"), instruct them to edit ",{"type":40,"tag":308,"props":698,"children":700},{"className":699},[],[701],{"type":46,"value":694},{"type":46,"value":703}," locally. Never accept, display, or commit API keys.",{"type":40,"tag":59,"props":705,"children":706},{},[707,712],{"type":40,"tag":63,"props":708,"children":709},{},[710],{"type":46,"value":711},"Execute, don't display:",{"type":46,"value":713}," When the agent supports terminal execution, run quick-start commands directly via the terminal tool rather than displaying bash blocks for the user to copy. Show bash blocks only when reviewing or when the user explicitly wants to copy commands.",{"type":40,"tag":59,"props":715,"children":716},{},[717,722],{"type":40,"tag":63,"props":718,"children":719},{},[720],{"type":46,"value":721},"Fine-tuning quick-start UX:",{"type":46,"value":723}," Brief the user before each command, confirm before continuing, and summarize completed vs. next actions.",{"type":40,"tag":59,"props":725,"children":726},{},[727,732,734,740,742,748,750,756,757,763,765,770],{"type":40,"tag":63,"props":728,"children":729},{},[730],{"type":46,"value":731},"When a user starts or completes fine-tuning, surface TensorBoard:",{"type":46,"value":733}," Remind them to inspect loss curves with ",{"type":40,"tag":308,"props":735,"children":737},{"className":736},[],[738],{"type":46,"value":739},"uv run tensorboard --logdir tb_logs\u002Ffinetuning",{"type":46,"value":741}," and open ",{"type":40,"tag":308,"props":743,"children":745},{"className":744},[],[746],{"type":46,"value":747},"http:\u002F\u002Flocalhost:6006",{"type":46,"value":749},". Look for steady descent in ",{"type":40,"tag":308,"props":751,"children":753},{"className":752},[],[754],{"type":46,"value":755},"train_loss",{"type":46,"value":288},{"type":40,"tag":308,"props":758,"children":760},{"className":759},[],[761],{"type":46,"value":762},"val_loss",{"type":46,"value":764},"; flag plateauing or divergence with concrete remediation prompts (see ",{"type":40,"tag":308,"props":766,"children":768},{"className":767},[],[769],{"type":46,"value":580},{"type":46,"value":771},").",{"type":40,"tag":59,"props":773,"children":774},{},[775,780,782,788,790,795],{"type":40,"tag":63,"props":776,"children":777},{},[778],{"type":46,"value":779},"After any code change in the fine-tuning workflow:",{"type":46,"value":781}," Run ",{"type":40,"tag":308,"props":783,"children":785},{"className":784},[],[786],{"type":46,"value":787},"uv run pytest",{"type":46,"value":789}," (or a targeted eval) before declaring success. Flag NaN, divergence, or plateauing loss with concrete remediation prompts (lower learning rate, gradient clipping, see ",{"type":40,"tag":308,"props":791,"children":793},{"className":792},[],[794],{"type":46,"value":580},{"type":46,"value":771},{"type":40,"tag":41,"props":797,"children":799},{"id":798},"routing",[800],{"type":46,"value":801},"Routing",{"type":40,"tag":49,"props":803,"children":804},{},[805,807,813,815],{"type":46,"value":806},"Format: ",{"type":40,"tag":308,"props":808,"children":810},{"className":809},[],[811],{"type":46,"value":812},"{{Doc}}",{"type":46,"value":814}," — ",{"type":40,"tag":816,"props":817,"children":819},"binding",{"value":818},"when to load",[],{"type":40,"tag":78,"props":821,"children":822},{},[823,832,841,850,859,868,877,886,903,912,922,931,940,950],{"type":40,"tag":59,"props":824,"children":825},{},[826,830],{"type":40,"tag":87,"props":827,"children":828},{"href":132},[829],{"type":46,"value":132},{"type":46,"value":831}," — User asks what Aurora is, how it works, why it matters, or needs background.",{"type":40,"tag":59,"props":833,"children":834},{},[835,839],{"type":40,"tag":87,"props":836,"children":837},{"href":89},[838],{"type":46,"value":89},{"type":46,"value":840}," — User asks why fine-tune Aurora, who fine-tuning is for, or what it unlocks.",{"type":40,"tag":59,"props":842,"children":843},{},[844,848],{"type":40,"tag":87,"props":845,"children":846},{"href":203},[847],{"type":46,"value":203},{"type":46,"value":849}," — User wants to see Aurora outputs interactively without training (best for weather\u002Fclimate practitioners).",{"type":40,"tag":59,"props":851,"children":852},{},[853,857],{"type":40,"tag":87,"props":854,"children":855},{"href":138},[856],{"type":46,"value":138},{"type":46,"value":858}," — User wants to run the Norway forecast or visualize Aurora predictions (best for non-technical audiences and first-look demos).",{"type":40,"tag":59,"props":860,"children":861},{},[862,866],{"type":40,"tag":87,"props":863,"children":864},{"href":96},[865],{"type":46,"value":96},{"type":46,"value":867}," — User wants to run a first fine-tuning experiment on bundled data.",{"type":40,"tag":59,"props":869,"children":870},{},[871,875],{"type":40,"tag":87,"props":872,"children":873},{"href":144},[874],{"type":46,"value":144},{"type":46,"value":876}," — User wants a new region for the inference example, custom prototype, or domain recipes.",{"type":40,"tag":59,"props":878,"children":879},{},[880,884],{"type":40,"tag":87,"props":881,"children":882},{"href":102},[883],{"type":46,"value":102},{"type":46,"value":885}," — Training\u002Foptimization, gradients, AMP, LoRA, variable extension.",{"type":40,"tag":59,"props":887,"children":888},{},[889,893,895,901],{"type":40,"tag":87,"props":890,"children":891},{"href":539},[892],{"type":46,"value":539},{"type":46,"value":894}," — Data format \u002F ",{"type":40,"tag":308,"props":896,"children":898},{"className":897},[],[899],{"type":46,"value":900},"aurora.Batch",{"type":46,"value":902}," structure.",{"type":40,"tag":59,"props":904,"children":905},{},[906,910],{"type":40,"tag":87,"props":907,"children":908},{"href":546},[909],{"type":46,"value":546},{"type":46,"value":911}," — Checkpoint selection.",{"type":40,"tag":59,"props":913,"children":914},{},[915,920],{"type":40,"tag":87,"props":916,"children":918},{"href":917},"docs\u002Finference-basics.md",[919],{"type":46,"value":917},{"type":46,"value":921}," — Running predictions without the Norway frontend.",{"type":40,"tag":59,"props":923,"children":924},{},[925,929],{"type":40,"tag":87,"props":926,"children":927},{"href":449},[928],{"type":46,"value":449},{"type":46,"value":930}," — Grid constraints, data formats, variable naming, deployment, validation.",{"type":40,"tag":59,"props":932,"children":933},{},[934,938],{"type":40,"tag":87,"props":935,"children":936},{"href":580},[937],{"type":46,"value":580},{"type":46,"value":939}," — Errors and diagnostics for both inference and fine-tuning.",{"type":40,"tag":59,"props":941,"children":942},{},[943,948],{"type":40,"tag":87,"props":944,"children":946},{"href":945},"starter-code\u002FREADME.md",[947],{"type":46,"value":945},{"type":46,"value":949}," — CLI and starter-code reference.",{"type":40,"tag":59,"props":951,"children":952},{},[953,958],{"type":40,"tag":87,"props":954,"children":956},{"href":955},"assets\u002Finference\u002Fscripts\u002FREADME.md",[957],{"type":46,"value":955},{"type":46,"value":959}," — Inference scripts and utilities.",{"type":40,"tag":586,"props":961,"children":963},{"id":962},"assets",[964],{"type":46,"value":965},"Assets",{"type":40,"tag":78,"props":967,"children":968},{},[969,980,991,1002,1013,1031],{"type":40,"tag":59,"props":970,"children":971},{},[972,978],{"type":40,"tag":308,"props":973,"children":975},{"className":974},[],[976],{"type":46,"value":977},"assets\u002Finference\u002Fnorway-example\u002F",{"type":46,"value":979}," — Reference inference implementation with frontend.",{"type":40,"tag":59,"props":981,"children":982},{},[983,989],{"type":40,"tag":308,"props":984,"children":986},{"className":985},[],[987],{"type":46,"value":988},"assets\u002Finference\u002Fscripts\u002F",{"type":46,"value":990}," — ERA5 download, validation, regional setup utilities.",{"type":40,"tag":59,"props":992,"children":993},{},[994,1000],{"type":40,"tag":308,"props":995,"children":997},{"className":996},[],[998],{"type":46,"value":999},"assets\u002Finference\u002Fpaper\u002F",{"type":46,"value":1001}," — Paper artifacts and supplementary material.",{"type":40,"tag":59,"props":1003,"children":1004},{},[1005,1011],{"type":40,"tag":308,"props":1006,"children":1008},{"className":1007},[],[1009],{"type":46,"value":1010},"assets\u002Ffinetune\u002Ffinetune-exploration-demo-app\u002F",{"type":46,"value":1012}," — Dockerized exploration demo (no training required).",{"type":40,"tag":59,"props":1014,"children":1015},{},[1016,1021,1023,1029],{"type":40,"tag":308,"props":1017,"children":1019},{"className":1018},[],[1020],{"type":46,"value":491},{"type":46,"value":1022}," — Full Python project (",{"type":40,"tag":308,"props":1024,"children":1026},{"className":1025},[],[1027],{"type":46,"value":1028},"vibe-tune-aurora",{"type":46,"value":1030},") with training loop, losses, data loaders, CLIs, and tests.",{"type":40,"tag":59,"props":1032,"children":1033},{},[1034,1039,1041,1046,1048,1053],{"type":40,"tag":308,"props":1035,"children":1037},{"className":1036},[],[1038],{"type":46,"value":475},{"type":46,"value":1040}," — Copies ",{"type":40,"tag":308,"props":1042,"children":1044},{"className":1043},[],[1045],{"type":46,"value":491},{"type":46,"value":1047}," into the user's repo as ",{"type":40,"tag":308,"props":1049,"children":1051},{"className":1050},[],[1052],{"type":46,"value":499},{"type":46,"value":106},{"type":40,"tag":41,"props":1055,"children":1057},{"id":1056},"reference-links",[1058],{"type":46,"value":1059},"Reference Links",{"type":40,"tag":78,"props":1061,"children":1062},{},[1063,1077,1089,1101],{"type":40,"tag":59,"props":1064,"children":1065},{},[1066,1068,1075],{"type":46,"value":1067},"Research Paper: ",{"type":40,"tag":87,"props":1069,"children":1072},{"href":1070,"rel":1071},"https:\u002F\u002Fwww.nature.com\u002Farticles\u002Fs41586-025-09005-y",[283],[1073],{"type":46,"value":1074},"A foundation model for the Earth system",{"type":46,"value":1076}," (Nature, 2025)",{"type":40,"tag":59,"props":1078,"children":1079},{},[1080,1082],{"type":46,"value":1081},"GitHub: ",{"type":40,"tag":87,"props":1083,"children":1086},{"href":1084,"rel":1085},"https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Faurora",[283],[1087],{"type":46,"value":1088},"github.com\u002Fmicrosoft\u002Faurora",{"type":40,"tag":59,"props":1090,"children":1091},{},[1092,1094],{"type":46,"value":1093},"Model Weights: ",{"type":40,"tag":87,"props":1095,"children":1098},{"href":1096,"rel":1097},"https:\u002F\u002Fhuggingface.co\u002Fmicrosoft\u002Faurora",[283],[1099],{"type":46,"value":1100},"huggingface.co\u002Fmicrosoft\u002Faurora",{"type":40,"tag":59,"props":1102,"children":1103},{},[1104,1106],{"type":46,"value":1105},"Azure AI Foundry: ",{"type":40,"tag":87,"props":1107,"children":1110},{"href":1108,"rel":1109},"https:\u002F\u002Fai.azure.com\u002Fcatalog\u002Fmodels\u002FAurora",[283],[1111],{"type":46,"value":1112},"ai.azure.com\u002Fcatalog\u002Fmodels\u002FAurora",{"items":1114,"total":1152},[1115,1122,1139],{"slug":4,"name":4,"fn":5,"description":6,"org":1116,"tags":1117,"stars":23,"repoUrl":24,"updatedAt":25},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1118,1119,1120,1121],{"name":17,"slug":18,"type":15},{"name":9,"slug":8,"type":15},{"name":21,"slug":22,"type":15},{"name":13,"slug":14,"type":15},{"slug":1123,"name":1123,"fn":1124,"description":1125,"org":1126,"tags":1127,"stars":23,"repoUrl":24,"updatedAt":1138},"msresearch-bioemu","generate protein conformational ensembles with BioEmu","Microsoft Research's BioEmu — generates protein conformational ensembles from amino acid sequence on a local GPU. Use when users mention BioEmu by name, want to predict protein dynamics or flexibility from sequence, or are setting up \u002F running \u002F troubleshooting BioEmu locally.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1128,1131,1134,1137],{"name":1129,"slug":1130,"type":15},"Bioinformatics","bioinformatics",{"name":1132,"slug":1133,"type":15},"Deep Learning","deep-learning",{"name":1135,"slug":1136,"type":15},"Life Sciences","life-sciences",{"name":9,"slug":8,"type":15},"2026-07-07T06:53:46.003518",{"slug":1140,"name":1140,"fn":1141,"description":1142,"org":1143,"tags":1144,"stars":23,"repoUrl":24,"updatedAt":1151},"msresearch-dayhoff","generate and score protein sequences with Dayhoff","Microsoft Research's Dayhoff — generates novel protein sequences and scores variants using hybrid Mamba+Transformer models trained on the 3.34B-sequence Dayhoff Atlas. Use when users mention Dayhoff by name, want to generate protein sequences from scratch or from a prompt, score mutations \u002F variants \u002F fitness without experimental data, design sequences in a protein family from homologs, scaffold sequences around a motif, or are setting up \u002F running \u002F troubleshooting the Dayhoff prototype locally.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1145,1146,1147,1148],{"name":1129,"slug":1130,"type":15},{"name":1132,"slug":1133,"type":15},{"name":9,"slug":8,"type":15},{"name":1149,"slug":1150,"type":15},"Proteomics","proteomics","2026-07-07T06:53:47.296995",3,{"items":1154,"total":1347},[1155,1177,1196,1217,1232,1249,1260,1273,1288,1303,1322,1335],{"slug":1156,"name":1156,"fn":1157,"description":1158,"org":1159,"tags":1160,"stars":1174,"repoUrl":1175,"updatedAt":1176},"rushstack-best-practices","manage Rush monorepos with best practices","Provides best practices and guidance for working with Rush monorepos. Use when the user is working in a Rush-based repository, asks about Rush commands (install, update, build, rebuild), needs help with project selection, dependency management, build caching, subspace configuration, or troubleshooting Rush-specific issues.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1161,1164,1167,1168,1171],{"name":1162,"slug":1163,"type":15},"Engineering","engineering",{"name":1165,"slug":1166,"type":15},"Local Development","local-development",{"name":9,"slug":8,"type":15},{"name":1169,"slug":1170,"type":15},"Project Management","project-management",{"name":1172,"slug":1173,"type":15},"Rush","rush",6484,"https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Frushstack","2026-04-06T18:34:44.965032",{"slug":1178,"name":1178,"fn":1179,"description":1180,"org":1181,"tags":1182,"stars":1193,"repoUrl":1194,"updatedAt":1195},"azure-ai-agents-persistent-dotnet","build AI agents with Azure .NET SDK","Azure AI Agents Persistent SDK for .NET. Low-level SDK for creating and managing AI agents with threads, messages, runs, and tools. Use for agent CRUD, conversation threads, streaming responses, function calling, file search, and code interpreter. Triggers: \"PersistentAgentsClient\", \"persistent agents\", \"agent threads\", \"agent runs\", \"streaming agents\", \"function calling agents .NET\".\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1183,1186,1189,1192],{"name":1184,"slug":1185,"type":15},".NET","net",{"name":1187,"slug":1188,"type":15},"Agents","agents",{"name":1190,"slug":1191,"type":15},"Azure","azure",{"name":17,"slug":18,"type":15},2804,"https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Fskills","2026-07-03T16:32:10.297433",{"slug":1197,"name":1197,"fn":1198,"description":1199,"org":1200,"tags":1201,"stars":1193,"repoUrl":1194,"updatedAt":1216},"azure-ai-anomalydetector-java","build anomaly detection applications with Java","Build anomaly detection applications with Azure AI Anomaly Detector SDK for Java. Use when implementing univariate\u002Fmultivariate anomaly detection, time-series analysis, or AI-powered monitoring.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1202,1205,1206,1209,1212,1213],{"name":1203,"slug":1204,"type":15},"Analytics","analytics",{"name":1190,"slug":1191,"type":15},{"name":1207,"slug":1208,"type":15},"Data Analysis","data-analysis",{"name":1210,"slug":1211,"type":15},"Java","java",{"name":9,"slug":8,"type":15},{"name":1214,"slug":1215,"type":15},"Monitoring","monitoring","2026-05-13T06:14:16.261754",{"slug":1218,"name":1218,"fn":1219,"description":1220,"org":1221,"tags":1222,"stars":1193,"repoUrl":1194,"updatedAt":1231},"azure-ai-contentsafety-java","build content moderation applications with Azure AI","Build content moderation applications with Azure AI Content Safety SDK for Java. Use when implementing text\u002Fimage analysis, blocklist management, or harm detection for hate, violence, sexual content, and self-harm.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1223,1226,1227,1228],{"name":1224,"slug":1225,"type":15},"AI Infrastructure","ai-infrastructure",{"name":1190,"slug":1191,"type":15},{"name":1210,"slug":1211,"type":15},{"name":1229,"slug":1230,"type":15},"Security","security","2026-07-07T06:53:31.293235",{"slug":1233,"name":1233,"fn":1234,"description":1235,"org":1236,"tags":1237,"stars":1193,"repoUrl":1194,"updatedAt":1248},"azure-ai-contentsafety-py","detect harmful content with Azure AI Content Safety","Azure AI Content Safety SDK for Python. Use for detecting harmful content in text and images with multi-severity classification.\nTriggers: \"azure-ai-contentsafety\", \"ContentSafetyClient\", \"content moderation\", \"harmful content\", \"text analysis\", \"image analysis\".\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1238,1239,1242,1243,1244,1247],{"name":1190,"slug":1191,"type":15},{"name":1240,"slug":1241,"type":15},"Compliance","compliance",{"name":17,"slug":18,"type":15},{"name":9,"slug":8,"type":15},{"name":1245,"slug":1246,"type":15},"Python","python",{"name":1229,"slug":1230,"type":15},"2026-07-18T05:14:23.017504",{"slug":1250,"name":1250,"fn":1251,"description":1252,"org":1253,"tags":1254,"stars":1193,"repoUrl":1194,"updatedAt":1259},"azure-ai-language-conversations-py","implement conversational language understanding with Python","Implement Conversational Language Understanding (CLU) using the azure-ai-language-conversations Python SDK. Use when working with ConversationAnalysisClient to analyze conversation intent and entities, building NLP features, or integrating language understanding into applications.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1255,1256,1257,1258],{"name":1203,"slug":1204,"type":15},{"name":1190,"slug":1191,"type":15},{"name":17,"slug":18,"type":15},{"name":1245,"slug":1246,"type":15},"2026-07-31T05:54:29.068751",{"slug":1261,"name":1261,"fn":1262,"description":1263,"org":1264,"tags":1265,"stars":1193,"repoUrl":1194,"updatedAt":1272},"azure-ai-translation-text-py","translate text using Azure AI services","Azure AI Text Translation SDK for real-time text translation, transliteration, language detection, and dictionary lookup. Use for translating text content in applications.\nTriggers: \"text translation\", \"translator\", \"translate text\", \"transliterate\", \"TextTranslationClient\".\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1266,1269,1270,1271],{"name":1267,"slug":1268,"type":15},"API Development","api-development",{"name":1190,"slug":1191,"type":15},{"name":9,"slug":8,"type":15},{"name":1245,"slug":1246,"type":15},"2026-07-18T05:14:16.988376",{"slug":1274,"name":1274,"fn":1275,"description":1276,"org":1277,"tags":1278,"stars":1193,"repoUrl":1194,"updatedAt":1287},"azure-ai-vision-imageanalysis-py","analyze images with Azure AI Vision","Azure AI Vision Image Analysis SDK for captions, tags, objects, OCR, people detection, and smart cropping. Use for computer vision and image understanding tasks.\nTriggers: \"image analysis\", \"computer vision\", \"OCR\", \"object detection\", \"ImageAnalysisClient\", \"image caption\".\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1279,1280,1283,1286],{"name":1190,"slug":1191,"type":15},{"name":1281,"slug":1282,"type":15},"Computer Vision","computer-vision",{"name":1284,"slug":1285,"type":15},"Images","images",{"name":1245,"slug":1246,"type":15},"2026-07-18T05:14:18.007737",{"slug":1289,"name":1289,"fn":1290,"description":1291,"org":1292,"tags":1293,"stars":1193,"repoUrl":1194,"updatedAt":1302},"azure-appconfiguration-java","manage configuration with Azure App Configuration","Azure App Configuration SDK for Java. Centralized application configuration management with key-value settings, feature flags, and snapshots.\nTriggers: \"ConfigurationClient java\", \"app configuration java\", \"feature flag java\", \"configuration setting java\", \"azure config java\".\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1294,1295,1298,1301],{"name":1190,"slug":1191,"type":15},{"name":1296,"slug":1297,"type":15},"Configuration","configuration",{"name":1299,"slug":1300,"type":15},"Feature Flags","feature-flags",{"name":1210,"slug":1211,"type":15},"2026-07-03T16:32:01.278468",{"slug":1304,"name":1304,"fn":1305,"description":1306,"org":1307,"tags":1308,"stars":1193,"repoUrl":1194,"updatedAt":1321},"azure-cosmos-rust","build applications with Azure Cosmos DB","Azure Cosmos DB library for Rust (NoSQL API). Document CRUD, containers, and globally distributed data.\nTriggers: \"cosmos db rust\", \"CosmosClient rust\", \"document crud rust\", \"NoSQL rust\", \"partition key rust\".\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1309,1312,1315,1318],{"name":1310,"slug":1311,"type":15},"Cosmos DB","cosmos-db",{"name":1313,"slug":1314,"type":15},"Database","database",{"name":1316,"slug":1317,"type":15},"NoSQL","nosql",{"name":1319,"slug":1320,"type":15},"Rust","rust","2026-07-31T05:54:27.021432",{"slug":1323,"name":1323,"fn":1305,"description":1324,"org":1325,"tags":1326,"stars":1193,"repoUrl":1194,"updatedAt":1334},"azure-cosmos-ts","Azure Cosmos DB JavaScript\u002FTypeScript SDK (@azure\u002Fcosmos) for data plane operations. Use for CRUD operations on documents, queries, bulk operations, and container management. Triggers: \"Cosmos DB\", \"@azure\u002Fcosmos\", \"CosmosClient\", \"document CRUD\", \"NoSQL queries\", \"bulk operations\", \"partition key\", \"container.items\".\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1327,1328,1329,1330,1331],{"name":1310,"slug":1311,"type":15},{"name":1313,"slug":1314,"type":15},{"name":9,"slug":8,"type":15},{"name":1316,"slug":1317,"type":15},{"name":1332,"slug":1333,"type":15},"TypeScript","typescript","2026-07-03T16:31:19.368382",{"slug":1336,"name":1336,"fn":1337,"description":1338,"org":1339,"tags":1340,"stars":1193,"repoUrl":1194,"updatedAt":1346},"azure-data-tables-java","build table storage applications with Java","Build table storage applications with Azure Tables SDK for Java. Use when working with Azure Table Storage or Cosmos DB Table API for NoSQL key-value data, schemaless storage, or structured data at scale.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[1341,1342,1343,1344,1345],{"name":1190,"slug":1191,"type":15},{"name":1310,"slug":1311,"type":15},{"name":1313,"slug":1314,"type":15},{"name":1210,"slug":1211,"type":15},{"name":1316,"slug":1317,"type":15},"2026-05-13T06:14:17.582229",267]