Update to NVIDIA BioNeMo Agent Toolkit

{
  "name": "rfdiffusion-nim",
  "description": "Run RFDiffusion protein backbone design via NVIDIA NIM. Use for de novo protein backbones, motif scaffolding, binder design, hotspot residues, contigs syntax, diffusion steps, hosted NVIDIA API calls, local Docker deployment, and PDB backbone outputs for ProteinMPNN sequence design.",
  "included_files": [
    {
      "relative_path": "references/api.md",
      "size_in_bytes": 4093
    },
    {
      "relative_path": "references/examples.md",
      "size_in_bytes": 857
    },
    {
      "relative_path": "references/parameters.md",
      "size_in_bytes": 899
    },
    {
      "relative_path": "references/science.md",
      "size_in_bytes": 975
    },
    {
      "relative_path": "references/validation.md",
      "size_in_bytes": 739
    }
  ],
  "skill_md_contents": "---\nname: rfdiffusion-nim\ndescription: >\n  Run RFDiffusion protein backbone design via NVIDIA NIM. Use for de novo protein backbones, motif scaffolding, binder design, hotspot residues, contigs syntax, diffusion steps, hosted NVIDIA API calls, local Docker deployment, and PDB backbone outputs for ProteinMPNN sequence design.\nlicense: Apache-2.0 AND CC-BY-4.0\ncompatibility: \"requests>=2.28\"\nallowed-tools: Bash, Read, Write, AskUserQuestion\n---\n\n# RFDiffusion NIM\n\nDesign protein backbone PDBs for de novo proteins, motif scaffolds, and binders.\nUse this `SKILL.md` for first-pass hosted/local usage; load supplemental files\nonly when needed:\n\n- `references/api.md`: exact endpoints, schemas, Docker flags, response fields.\n- `references/science.md`: design modes, strengths, limits, and handoffs.\n- `references/parameters.md`: contigs, hotspots, steps, and seeds.\n- `references/validation.md`: PDB, contig, and artifact sanity checks.\n- `references/examples.md`: compact hosted/local request patterns.\n\n## Choose Mode\n\nAsk only when context is unclear:\n\n> Hosted NVIDIA API or local Docker NIM?\n\n- Hosted: `https://health.api.nvidia.com/v1/biology/ipd/rfdiffusion/generate`\n- Local: `http://localhost:8000/biology/ipd/rfdiffusion/generate`\n\nLocal inference paths do not include `/v1/`. Hosted requests use `Authorization: Bearer $NGC_API_KEY`. Supported local Docker\nstartup uses `NGC_API_KEY` (or `NVIDIA_API_KEY` via the preflight) for\nregistry login, entitlement checks, and first-run model downloads; pass it\ninto the container with `-e NGC_API_KEY`. Local inference requests use no\nauth header after readiness. Warm-cache key-free startup varies by\nimage/version and should not be assumed.\n\n## Local Docker\n\nFor local setup answers, copy the preflight below exactly before `docker login`,\n`docker run`, readiness, and the no-auth local request. Do not replace it with a\nsimple `: \"${NGC_API_KEY:?Set NGC_API_KEY}\"` check, do not invent a cache\ndefault, and do not drop the `NVIDIA_API_KEY` fallback. Default setup is single\nGPU `device=0`.\n\n```bash\nset -a\n[ -f .env ] && . ./.env\nset +a\n\nif [ -z \"${NGC_API_KEY:-}\" ] && [ -n \"${NVIDIA_API_KEY:-}\" ]; then\n  export NGC_API_KEY=\"$NVIDIA_API_KEY\"\nfi\n: \"${NGC_API_KEY:?Set NGC_API_KEY or NVIDIA_API_KEY}\"\n: \"${LOCAL_NIM_CACHE:?Set LOCAL_NIM_CACHE}\"\n\necho \"$NGC_API_KEY\" | docker login nvcr.io --username '$oauthtoken' --password-stdin\n\nmkdir -p \"${LOCAL_NIM_CACHE}\"\nchmod 777 \"${LOCAL_NIM_CACHE}\"\n\ndocker run -it \\\n  --runtime=nvidia \\\n  --gpus \"device=0\" \\\n  -e NGC_API_KEY \\\n  -v \"${LOCAL_NIM_CACHE}:/opt/nim/.cache\" \\\n  -p 8000:8000 \\\n  nvcr.io/nim/ipd/rfdiffusion:2\n```\n\nReadiness:\n\n```bash\nuntil curl -sf http://localhost:8000/v1/health/ready; do sleep 5; done\n```\n\n## Contigs DSL\n\n`contigs` defines what to keep and what to generate.\n\n- `\"100\"`: generate exactly 100 residues.\n- `\"80-120\"`: generate 80-120 residues.\n- `\"A25-35\"`: keep chain A residues 25-35 from `input_pdb`.\n- `\"A25-35/0 50-80\"`: keep A25-35, insert chain break `/0`, generate 50-80.\n\nDesign modes:\n\n- De novo: `contigs=\"80-120\"`; live hosted validation requires a non-empty\n  `input_pdb` or `input_pdb_asset`, so inline requests should include the dummy\n  PDB below.\n- Motif scaffolding: read `target.pdb`, pass `input_pdb`, use a contig like\n  `\"A25-35/0 50-80\"`.\n- Binder design: pass target `input_pdb`, contig with target and binder segment,\n  and `hotspot_res=[\"A50\", \"A51\", ...]` in ChainResidue string format.\n\n```python\nDUMMY_PDB = (\n    \"CRYST1    1.000    1.000    1.000  90.00  90.00  90.00 P 1           1\\n\"\n    \"ATOM      1  CA  ALA A   1       0.000   0.000   0.000  1.00  0.00           C\\n\"\n    \"END\\n\"\n)\n```\n\n## Request Pattern\n\n```python\nimport os\nfrom pathlib import Path\nimport requests\n\nHOSTED = True\nurl = (\n    \"https://health.api.nvidia.com/v1/biology/ipd/rfdiffusion/generate\"\n    if HOSTED else \"http://localhost:8000/biology/ipd/rfdiffusion/generate\"\n)\nheaders = {\"Content-Type\": \"application/json\"}\nif HOSTED:\n    headers[\"Authorization\"] = f\"Bearer {os.environ['NGC_API_KEY']}\"\n\npayload = {\n    \"input_pdb\": DUMMY_PDB,\n    \"contigs\": \"80-120\",\n    \"diffusion_steps\": 50,\n}\nresponse = requests.post(url, headers=headers, json=payload, timeout=300)\nresponse.raise_for_status()\nresult = response.json()\nPath(\"designed_backbone.pdb\").write_text(result[\"output_pdb\"])\n```\n\nMotif scaffold:\n\n```python\npayload = {\n    \"input_pdb\": Path(\"target.pdb\").read_text(),\n    \"contigs\": \"A25-35/0 50-80\",\n    \"diffusion_steps\": 50,\n}\n```\n\nBinder design:\n\n```python\npayload = {\n    \"input_pdb\": Path(\"target.pdb\").read_text(),\n    \"contigs\": \"A1-100/0 50-100\",\n    \"hotspot_res\": [\"A50\", \"A51\", \"A52\", \"A53\", \"A54\"],\n    \"diffusion_steps\": 50,\n}\n```\n\n## Save And Interpret Output\n\nSave `result[\"output_pdb\"]` as a PDB artifact and report `elapsed_ms` when\npresent. Generated backbones are not final proteins; feed them to ProteinMPNN\nfor sequence design, then validate sequences/structures with Boltz2 or\nOpenFold3. For PDB and contig checks, read `references/validation.md`.\n\n## Limits And Troubleshooting\n\n- `diffusion_steps`: 1-50; 50 is maximum quality, fewer is faster.\n- Single GPU; minimum GPU VRAM is about 12 GB.\n- `hotspot_res` uses strings like `\"A50\"`, not tuples.\n- `422` usually means chain IDs in `contigs`/`hotspot_res` do not match\n  `input_pdb`, a malformed contig, or omitted `input_pdb` for hosted de novo.\n- Local URL 404 usually means an accidental `/v1/` prefix.\n"
}