[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6QGP7XAD4NNZ1RBB97M7A7T":3,"skill-same:01M6QGP7XAD4NNZ1RBB97M7A7T":106},{"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":38,"linkOnly":39,"markdownContent":40,"markdownBody":41,"frontMatter":42,"refName":51,"commitSha":52,"skillDirectory":53,"forksCount":14,"repositoryHtmlUrl":54,"repositoryHomepageUrl":55,"ownerHtmlUrl":56,"ownerBlogUrl":57,"ownerDescription":58,"updatedAt":59,"install":60,"resources":73},"01M6QGP7XAD4NNZ1RBB97M7A7T","esm",null,"Uses the Biohub esm Python SDK for ESM3 protein generation, ESMC embeddings, and ESMFold2 all-atom folding. Applies to local model inference and Biohub hosted clients, including former Forge workflows; distinguishes the separate legacy fair-esm distribution.","Tashima-Tarsh\u002FDisha6.6","Tashima-Tarsh","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F269906498?v=4","01M6PBH98X03XERH0HFT4PFB1P","NITISH KUMAR","01M6PBH9DA0M6KM3Z12P848F0K",0,"MIT","https:\u002F\u002Fgithub.com\u002FTashima-Tarsh\u002FDisha6.6\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fkd-esm\u002FSKILL.md",[],[5,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37],"model","for","scientific","atom","distribution","local","Legacy","use","environment","python sdk","python","access","api","forge","network","embeddings","gpu","workflows","memory","2026-11-10 10:12:30",false,"---\nname: esm\ndescription: Uses the Biohub esm Python SDK for ESM3 protein generation, ESMC embeddings, and ESMFold2 all-atom folding. Applies to local model inference and Biohub hosted clients, including former Forge workflows; distinguishes the separate legacy fair-esm distribution.\nlicense: MIT license\ncompatibility: Requires Python 3.12+ and esm 3.4.1.post1. Local pretrained inference needs model weights and sufficient RAM or GPU memory; hosted inference needs network access and ESM_API_KEY. Use an isolated environment, separate from fair-esm.\nmetadata:\n  version: \"2.0\"\n  skill-author: K-Dense Inc.\n  last-reviewed: \"2026-09-30\"\n  upstream-version: \"3.4.1.post1\"\ncategory: \"scientific-computing\"\n---\n\n# ESM: protein generation, embeddings, and folding\n\n## When to use\n\nUse for Biohub's `esm` SDK: ESM3 masked multimodal generation, ESMC sequence\nrepresentations, and ESMFold2 all-atom prediction. The former Forge platform\nmigrated to `https:\u002F\u002Fbiohub.ai`, including ESM3; SDK class names still contain\n`Forge`. This skill targets the released **esm 3.4.1.post1**, verified against\nits wheel and current official documentation.\n\n`fair-esm` is Meta's separate, older ESM2\u002FESMFold\u002FESM-IF distribution. Both\npackages import as `esm`; install them in separate environments. The legacy\n`esm.pretrained.esm2_*` interface is not the Biohub ESMC interface. Hugging Face\nTransformers' native ESMC implementation is another API: do not interchange its\nversion requirements or output types with this SDK.\n\n## Setup and model choice\n\n```bash\nuv venv --python 3.12 .venv-esm\nuv pip install --python .venv-esm\u002Fbin\u002Fpython \"esm==3.4.1.post1\"\n```\n\nThe release declares Python >=3.12, Torch >=2.11,\u003C2.12 and Transformers\n>=4.57.6,\u003C5. Python 3.12, Torch 2.11.0 and Transformers 4.57.6 were tested on\nCPU. Linux x86_64 installs include GPU-specific dependencies; use a supported\ninference machine and budget disk space. Flash Attention is optional; it is\nunnecessary for the tiny CPU tests. Do not install this stack into a shared\nenvironment containing incompatible Transformers or Torch pins.\n\n| Task | Local model\u002FAPI | Hosted client\u002Fmodel |\n| --- | --- | --- |\n| ESM3 sequence\u002Fstructure\u002Ffunction generation | `ESM3.from_pretrained(\"esm3-sm-open-v1\")` | `client(\"esm3-medium-2024-08\")`; small\u002Flarge IDs in the ESM3 reference |\n| ESMC embeddings | `EsmcForMaskedLM` and `EsmcTokenizer`; `biohub\u002FESMC-300M`, `biohub\u002FESMC-600M`, `biohub\u002FESMC-6B` | `esmc_client(\"esmc-600m-2024-12\")`; 300M\u002F6B also documented |\n| ESMFold2 all-atom structures | `EsmFold2Model`, `ESMFold2InputBuilder`; `biohub\u002FESMFold2` | `esmfold2_client(\"esmfold2-fast-2026-05\")` |\n\nESMC 6B weights are now available locally. Current Biohub model cards identify\nMIT licensing, with ESMC cards also linking third-party notices; review the\nexact selected artifact's card and access requirements.\nThe older `ESMC` class remains as a deprecated compatibility wrapper. Local\nmodel sizes, hosted availability and account quotas are different constraints;\na larger model does not guarantee better performance on a particular assay.\n\n## Workflow\n\n1. Specify the sequence\u002Fchain\u002Fresidue mapping and whether the objective is\n   generation, representation extraction or folding. Preserve input IDs and\n   source provenance. Reject accidental `...`, spaces, FASTA headers or gaps\n   in plain single-chain sequences rather than silently deleting them.\n2. Choose a local checkpoint or explicit hosted model ID. Record SDK, checkpoint\n   revision, model settings, mask locations and random seed where supported.\n3. Check every SDK result for `ESMProteinError`. Hosted failures may be returned\n   as values. Use finite request timeouts, context managers and bounded\n   concurrency. See [hosted contracts](references\u002Fforge-api.md).\n4. Validate the output contract: sequence length and fixed residues, residue-only\n   embedding axes, or chain\u002Fatom mapping and confidence. Keep model predictions\n   separate from experimental validation.\n5. Save sequences\u002Fstructures with settings and identifiers. Avoid caches keyed\n   only by sequence when model, structure, function or generation settings differ.\n\n## ESM3 completion and fresh structure prediction\n\nIllustrative pretrained inference; no weights or hosted jobs were run in this\nrefresh. This toy sequence demonstrates API mechanics, not a functional design.\n\n```python\nimport torch\nfrom esm.models.esm3 import ESM3\nfrom esm.sdk.api import ESMProtein, ESMProteinError, GenerationConfig\n\nmodel = ESM3.from_pretrained(\"esm3-sm-open-v1\", device=torch.device(\"cpu\"))\nprompt = \"MPRT___KEND\"\ncompleted = model.generate(\n    ESMProtein(sequence=prompt),\n    GenerationConfig(track=\"sequence\", num_steps=3, temperature=0.7),\n)\nif isinstance(completed, ESMProteinError):\n    raise completed\nassert completed.sequence is not None and len(completed.sequence) == len(prompt)\nassert \"_\" not in completed.sequence\nassert all(a == \"_\" or a == b for a, b in zip(prompt, completed.sequence))\n\n# A fresh sequence-only prompt prevents old coordinates conditioning the check.\nfolded = model.generate(\n    ESMProtein(sequence=completed.sequence),\n    GenerationConfig(track=\"structure\", num_steps=8),\n)\nif isinstance(folded, ESMProteinError):\n    raise folded\nassert folded.coordinates is not None\nfolded.to_pdb(\"candidate.pdb\")\n```\n\nGeneration fills masked positions. Calling it again on a completed track does\nnot implement refinement or temperature annealing; explicitly remask selected\npositions or clear the track. ESM3 structure generation and ESMFold2 prediction\nuse different models and result types. See [ESM3](references\u002Fesm3-api.md) for\ninverse folding, function vocabulary and coordinate conventions.\n\n## ESMC embeddings with correct residue pooling\n\nIllustrative pretrained loading; the same API and pooling were executed with a\ntiny randomly initialized model on CPU. Add this skill's `scripts\u002F` directory to\n`PYTHONPATH` when importing the bundled helper.\n\n```python\nfrom esm.models.esmc import EsmcForMaskedLM, EsmcTokenizer\nfrom esm_embeddings import embed_sequences\n\nmodel = EsmcForMaskedLM.from_pretrained(\"biohub\u002FESMC-300M\", device=\"cpu\").eval()\ntokenizer = EsmcTokenizer()\nsequences = [\"MPRTKEINDAGLIVHSPQWFYK\", \"ACDEFGHIK\"]\nfeatures = embed_sequences(model, tokenizer, sequences)\nassert features.shape == (2, 960)\n```\n\n`output.last_hidden_state` has shape `(B,T,D)` and includes CLS, EOS and padding;\n`T` is not the raw residue count. [scripts\u002Fesm_embeddings.py](scripts\u002Fesm_embeddings.py)\nperforms one real padded batch, excludes special\u002Fpadding tokens, validates the\nresidue count, and returns `(B,D)` CPU features in input order. Choose batch size\nby sequence lengths and available memory. It does not truncate, download weights\nor contact a service. See [ESMC](references\u002Fesm-c-api.md) for hosted output types,\nper-residue extraction, gradient behavior and migration details.\n\n## Hosted authentication and folding\n\nRead only the intended credential from the environment; pass it explicitly so\nit is resolved when the client is created. SDK factory defaults capture\n`ESM_API_KEY` at import time. Create keys in the\n[Biohub developer console](https:\u002F\u002Fbiohub.ai\u002Fdeveloper-console\u002Fapi-keys).\n\nIllustrative authenticated inference:\n\n```python\nimport os\nfrom esm.sdk import esmfold2_client\nfrom esm.sdk.api import ESMProteinError, FoldingConfig\nfrom esm.utils.structure.input_builder import ProteinInput, StructurePredictionInput\n\nfold_input = StructurePredictionInput(\n    sequences=[ProteinInput(id=\"A\", sequence=\"MPRTKEINDAGLIVHSPQWFYK\")]\n)\nwith esmfold2_client(\n    model=\"esmfold2-fast-2026-05\", url=\"https:\u002F\u002Fbiohub.ai\",\n    token=os.environ[\"ESM_API_KEY\"], request_timeout=300,\n) as client:\n    result = client.fold_all_atom(fold_input, config=FoldingConfig())\nif isinstance(result, ESMProteinError):\n    raise result\nwith open(\"candidate.cif\", \"w\") as handle:\n    handle.write(result.complex.to_mmcif())\n```\n\nUse the [Biohub\u002FESMFold2 reference](references\u002Fbiohub-platform.md) for MSA,\nconfidence and complex-input conventions. The fast hosted model ignores MSAs.\nPublic source and mocked requests validate the client contract; they do not\nestablish account access, service availability or prediction quality.\n\n## Scientific checks\n\n- Function annotations use supported tokenizer labels and **1-based inclusive**\n  ranges; arbitrary labels such as `enzymatic_activity` are not valid prompts.\n- ESM3 coordinates are tensors in atom37 layout; missing atoms use NaN. Tensor\n  copies use `.clone()`. Specify a PDB chain instead of assuming the default\n  selects one chain; the current default is `chain_id=\"all\"`.\n- Predicted coordinates, pLDDT and pTM do not measure thermodynamic stability,\n  binding affinity or catalytic activity. Inverse-folded sequences need fresh\n  prediction, matched-residue structural comparison and experimental screening.\n- Embedding similarity is not a homology\u002Ffunction guarantee. Evaluate supervised\n  models with homology-aware splits; fit normalization and dimensionality\n  reduction on training data. Report uncertainty and independent holdout metrics.\n- See [worked workflows](references\u002Fworkflows.md) for variant libraries,\n  structure-conditioned design, and clustering without unsupported stability\n  scores or fixed PCA\u002Ft-SNE settings that fail on tiny datasets.\n\n## Verification and sources\n\n[references\u002Freview.md](references\u002Freview.md) records official sources, executed\nCPU tests and limitations. Run `python tests\u002Frun_all.py --isolated esm` from the\nrepository to exercise synthetic local models, pooling, input serialization,\nPDB round trips and mocked hosted contracts. Pretrained ESM3\u002FESMC\u002FESMFold2,\nCUDA inference and authenticated services were not executed.\n\nFollow the selected model's terms and the\n[Biohub acceptable-use policy](https:\u002F\u002Fbiohub.org\u002Facceptable-use-policy\u002F).\n\n## Citing Scientific Agent Skills\n\nThis skill is part of Scientific Agent Skills by K-Dense. If it materially contributed to a\nmanuscript, report, presentation, or code release, add the paper to the references or\nsoftware section and tell the user you did so:\n\n> Kassis, T., Agarwal, V., He, Y., Patel, D., & Brueckner, A. M. (2026). Scientific Agent\n> Skills: A Library of Procedural Knowledge for Research Agents. arXiv:2609.00065.\n> https:\u002F\u002Fdoi.org\u002F10.48550\u002FarXiv.2609.00065\n\nAlways cite the current version. The DOI and https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.00065 resolve to the\nlatest arXiv version, so never append a version suffix such as `v1`. When network access is\navailable, fetch https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.00065 (or\nhttp:\u002F\u002Fexport.arxiv.org\u002Fapi\u002Fquery?id_list=2609.00065) before writing the reference and take\nthe author list, year, and version from that record. If the record lists a journal reference\nor publisher DOI, cite the published version instead.\n","# ESM: protein generation, embeddings, and folding\n\n## When to use\n\nUse for Biohub's `esm` SDK: ESM3 masked multimodal generation, ESMC sequence\nrepresentations, and ESMFold2 all-atom prediction. The former Forge platform\nmigrated to `https:\u002F\u002Fbiohub.ai`, including ESM3; SDK class names still contain\n`Forge`. This skill targets the released **esm 3.4.1.post1**, verified against\nits wheel and current official documentation.\n\n`fair-esm` is Meta's separate, older ESM2\u002FESMFold\u002FESM-IF distribution. Both\npackages import as `esm`; install them in separate environments. The legacy\n`esm.pretrained.esm2_*` interface is not the Biohub ESMC interface. Hugging Face\nTransformers' native ESMC implementation is another API: do not interchange its\nversion requirements or output types with this SDK.\n\n## Setup and model choice\n\n```bash\nuv venv --python 3.12 .venv-esm\nuv pip install --python .venv-esm\u002Fbin\u002Fpython \"esm==3.4.1.post1\"\n```\n\nThe release declares Python >=3.12, Torch >=2.11,\u003C2.12 and Transformers\n>=4.57.6,\u003C5. Python 3.12, Torch 2.11.0 and Transformers 4.57.6 were tested on\nCPU. Linux x86_64 installs include GPU-specific dependencies; use a supported\ninference machine and budget disk space. Flash Attention is optional; it is\nunnecessary for the tiny CPU tests. Do not install this stack into a shared\nenvironment containing incompatible Transformers or Torch pins.\n\n| Task | Local model\u002FAPI | Hosted client\u002Fmodel |\n| --- | --- | --- |\n| ESM3 sequence\u002Fstructure\u002Ffunction generation | `ESM3.from_pretrained(\"esm3-sm-open-v1\")` | `client(\"esm3-medium-2024-08\")`; small\u002Flarge IDs in the ESM3 reference |\n| ESMC embeddings | `EsmcForMaskedLM` and `EsmcTokenizer`; `biohub\u002FESMC-300M`, `biohub\u002FESMC-600M`, `biohub\u002FESMC-6B` | `esmc_client(\"esmc-600m-2024-12\")`; 300M\u002F6B also documented |\n| ESMFold2 all-atom structures | `EsmFold2Model`, `ESMFold2InputBuilder`; `biohub\u002FESMFold2` | `esmfold2_client(\"esmfold2-fast-2026-05\")` |\n\nESMC 6B weights are now available locally. Current Biohub model cards identify\nMIT licensing, with ESMC cards also linking third-party notices; review the\nexact selected artifact's card and access requirements.\nThe older `ESMC` class remains as a deprecated compatibility wrapper. Local\nmodel sizes, hosted availability and account quotas are different constraints;\na larger model does not guarantee better performance on a particular assay.\n\n## Workflow\n\n1. Specify the sequence\u002Fchain\u002Fresidue mapping and whether the objective is\n   generation, representation extraction or folding. Preserve input IDs and\n   source provenance. Reject accidental `...`, spaces, FASTA headers or gaps\n   in plain single-chain sequences rather than silently deleting them.\n2. Choose a local checkpoint or explicit hosted model ID. Record SDK, checkpoint\n   revision, model settings, mask locations and random seed where supported.\n3. Check every SDK result for `ESMProteinError`. Hosted failures may be returned\n   as values. Use finite request timeouts, context managers and bounded\n   concurrency. See [hosted contracts](references\u002Fforge-api.md).\n4. Validate the output contract: sequence length and fixed residues, residue-only\n   embedding axes, or chain\u002Fatom mapping and confidence. Keep model predictions\n   separate from experimental validation.\n5. Save sequences\u002Fstructures with settings and identifiers. Avoid caches keyed\n   only by sequence when model, structure, function or generation settings differ.\n\n## ESM3 completion and fresh structure prediction\n\nIllustrative pretrained inference; no weights or hosted jobs were run in this\nrefresh. This toy sequence demonstrates API mechanics, not a functional design.\n\n```python\nimport torch\nfrom esm.models.esm3 import ESM3\nfrom esm.sdk.api import ESMProtein, ESMProteinError, GenerationConfig\n\nmodel = ESM3.from_pretrained(\"esm3-sm-open-v1\", device=torch.device(\"cpu\"))\nprompt = \"MPRT___KEND\"\ncompleted = model.generate(\n    ESMProtein(sequence=prompt),\n    GenerationConfig(track=\"sequence\", num_steps=3, temperature=0.7),\n)\nif isinstance(completed, ESMProteinError):\n    raise completed\nassert completed.sequence is not None and len(completed.sequence) == len(prompt)\nassert \"_\" not in completed.sequence\nassert all(a == \"_\" or a == b for a, b in zip(prompt, completed.sequence))\n\n# A fresh sequence-only prompt prevents old coordinates conditioning the check.\nfolded = model.generate(\n    ESMProtein(sequence=completed.sequence),\n    GenerationConfig(track=\"structure\", num_steps=8),\n)\nif isinstance(folded, ESMProteinError):\n    raise folded\nassert folded.coordinates is not None\nfolded.to_pdb(\"candidate.pdb\")\n```\n\nGeneration fills masked positions. Calling it again on a completed track does\nnot implement refinement or temperature annealing; explicitly remask selected\npositions or clear the track. ESM3 structure generation and ESMFold2 prediction\nuse different models and result types. See [ESM3](references\u002Fesm3-api.md) for\ninverse folding, function vocabulary and coordinate conventions.\n\n## ESMC embeddings with correct residue pooling\n\nIllustrative pretrained loading; the same API and pooling were executed with a\ntiny randomly initialized model on CPU. Add this skill's `scripts\u002F` directory to\n`PYTHONPATH` when importing the bundled helper.\n\n```python\nfrom esm.models.esmc import EsmcForMaskedLM, EsmcTokenizer\nfrom esm_embeddings import embed_sequences\n\nmodel = EsmcForMaskedLM.from_pretrained(\"biohub\u002FESMC-300M\", device=\"cpu\").eval()\ntokenizer = EsmcTokenizer()\nsequences = [\"MPRTKEINDAGLIVHSPQWFYK\", \"ACDEFGHIK\"]\nfeatures = embed_sequences(model, tokenizer, sequences)\nassert features.shape == (2, 960)\n```\n\n`output.last_hidden_state` has shape `(B,T,D)` and includes CLS, EOS and padding;\n`T` is not the raw residue count. [scripts\u002Fesm_embeddings.py](scripts\u002Fesm_embeddings.py)\nperforms one real padded batch, excludes special\u002Fpadding tokens, validates the\nresidue count, and returns `(B,D)` CPU features in input order. Choose batch size\nby sequence lengths and available memory. It does not truncate, download weights\nor contact a service. See [ESMC](references\u002Fesm-c-api.md) for hosted output types,\nper-residue extraction, gradient behavior and migration details.\n\n## Hosted authentication and folding\n\nRead only the intended credential from the environment; pass it explicitly so\nit is resolved when the client is created. SDK factory defaults capture\n`ESM_API_KEY` at import time. Create keys in the\n[Biohub developer console](https:\u002F\u002Fbiohub.ai\u002Fdeveloper-console\u002Fapi-keys).\n\nIllustrative authenticated inference:\n\n```python\nimport os\nfrom esm.sdk import esmfold2_client\nfrom esm.sdk.api import ESMProteinError, FoldingConfig\nfrom esm.utils.structure.input_builder import ProteinInput, StructurePredictionInput\n\nfold_input = StructurePredictionInput(\n    sequences=[ProteinInput(id=\"A\", sequence=\"MPRTKEINDAGLIVHSPQWFYK\")]\n)\nwith esmfold2_client(\n    model=\"esmfold2-fast-2026-05\", url=\"https:\u002F\u002Fbiohub.ai\",\n    token=os.environ[\"ESM_API_KEY\"], request_timeout=300,\n) as client:\n    result = client.fold_all_atom(fold_input, config=FoldingConfig())\nif isinstance(result, ESMProteinError):\n    raise result\nwith open(\"candidate.cif\", \"w\") as handle:\n    handle.write(result.complex.to_mmcif())\n```\n\nUse the [Biohub\u002FESMFold2 reference](references\u002Fbiohub-platform.md) for MSA,\nconfidence and complex-input conventions. The fast hosted model ignores MSAs.\nPublic source and mocked requests validate the client contract; they do not\nestablish account access, service availability or prediction quality.\n\n## Scientific checks\n\n- Function annotations use supported tokenizer labels and **1-based inclusive**\n  ranges; arbitrary labels such as `enzymatic_activity` are not valid prompts.\n- ESM3 coordinates are tensors in atom37 layout; missing atoms use NaN. Tensor\n  copies use `.clone()`. Specify a PDB chain instead of assuming the default\n  selects one chain; the current default is `chain_id=\"all\"`.\n- Predicted coordinates, pLDDT and pTM do not measure thermodynamic stability,\n  binding affinity or catalytic activity. Inverse-folded sequences need fresh\n  prediction, matched-residue structural comparison and experimental screening.\n- Embedding similarity is not a homology\u002Ffunction guarantee. Evaluate supervised\n  models with homology-aware splits; fit normalization and dimensionality\n  reduction on training data. Report uncertainty and independent holdout metrics.\n- See [worked workflows](references\u002Fworkflows.md) for variant libraries,\n  structure-conditioned design, and clustering without unsupported stability\n  scores or fixed PCA\u002Ft-SNE settings that fail on tiny datasets.\n\n## Verification and sources\n\n[references\u002Freview.md](references\u002Freview.md) records official sources, executed\nCPU tests and limitations. Run `python tests\u002Frun_all.py --isolated esm` from the\nrepository to exercise synthetic local models, pooling, input serialization,\nPDB round trips and mocked hosted contracts. Pretrained ESM3\u002FESMC\u002FESMFold2,\nCUDA inference and authenticated services were not executed.\n\nFollow the selected model's terms and the\n[Biohub acceptable-use policy](https:\u002F\u002Fbiohub.org\u002Facceptable-use-policy\u002F).\n\n## Citing Scientific Agent Skills\n\nThis skill is part of Scientific Agent Skills by K-Dense. If it materially contributed to a\nmanuscript, report, presentation, or code release, add the paper to the references or\nsoftware section and tell the user you did so:\n\n> Kassis, T., Agarwal, V., He, Y., Patel, D., & Brueckner, A. M. (2026). Scientific Agent\n> Skills: A Library of Procedural Knowledge for Research Agents. arXiv:2609.00065.\n> https:\u002F\u002Fdoi.org\u002F10.48550\u002FarXiv.2609.00065\n\nAlways cite the current version. The DOI and https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.00065 resolve to the\nlatest arXiv version, so never append a version suffix such as `v1`. When network access is\navailable, fetch https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.00065 (or\nhttp:\u002F\u002Fexport.arxiv.org\u002Fapi\u002Fquery?id_list=2609.00065) before writing the reference and take\nthe author list, year, and version from that record. If the record lists a journal reference\nor publisher DOI, cite the published version instead.\n",{"name":5,"license":43,"category":44,"metadata":45,"description":7,"compatibility":50},"MIT license","scientific-computing",{"version":46,"skill-author":47,"last-reviewed":48,"upstream-version":49},"2.0","K-Dense Inc.","2026-09-30","3.4.1.post1","Requires Python 3.12+ and esm 3.4.1.post1. Local pretrained inference needs model weights and sufficient RAM or GPU memory; hosted inference needs network access and ESM_API_KEY. Use an isolated environment, separate from fair-esm.","main","0da4485785ee599388900be50db1635489d53056",".agents\u002Fskills\u002Fkd-esm","https:\u002F\u002Fgithub.com\u002FTashima-Tarsh\u002FDisha6.6","https:\u002F\u002Fdisha6.6.thenitishkr.in\u002F","https:\u002F\u002Fgithub.com\u002FTashima-Tarsh","www.thenitishkr","thenitishkr | AI • Cyber-security • Research\r\n","2026-11-08 03:33:25",{"source":8,"skillName":5,"commands":61,"prompt":71,"zipAvailable":72},[62,65,68],{"tool":63,"command":64},"npx","npx skills add Tashima-Tarsh\u002FDisha6.6 --skill esm",{"tool":66,"command":67},"bunx","bunx skills add Tashima-Tarsh\u002FDisha6.6 --skill esm",{"tool":69,"command":70},"pnpm","pnpm dlx skills add Tashima-Tarsh\u002FDisha6.6 --skill esm","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FTashima-Tarsh\u002FDisha6.6\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fkd-esm\u002FSKILL.md",true,[74,80,84,88,92,96,100],{"path":75,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":78,"sourceHtmlUrl":79},"references\u002Fbiohub-platform.md","REFERENCE","参考文档",5389,"https:\u002F\u002Fgithub.com\u002FTashima-Tarsh\u002FDisha6.6\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fkd-esm\u002Freferences\u002Fbiohub-platform.md",{"path":81,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":82,"sourceHtmlUrl":83},"references\u002Fesm-c-api.md",6655,"https:\u002F\u002Fgithub.com\u002FTashima-Tarsh\u002FDisha6.6\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fkd-esm\u002Freferences\u002Fesm-c-api.md",{"path":85,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":86,"sourceHtmlUrl":87},"references\u002Fesm3-api.md",6693,"https:\u002F\u002Fgithub.com\u002FTashima-Tarsh\u002FDisha6.6\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fkd-esm\u002Freferences\u002Fesm3-api.md",{"path":89,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":90,"sourceHtmlUrl":91},"references\u002Fforge-api.md",7029,"https:\u002F\u002Fgithub.com\u002FTashima-Tarsh\u002FDisha6.6\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fkd-esm\u002Freferences\u002Fforge-api.md",{"path":93,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":94,"sourceHtmlUrl":95},"references\u002Freview.md",6057,"https:\u002F\u002Fgithub.com\u002FTashima-Tarsh\u002FDisha6.6\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fkd-esm\u002Freferences\u002Freview.md",{"path":97,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":98,"sourceHtmlUrl":99},"references\u002Fworkflows.md",7248,"https:\u002F\u002Fgithub.com\u002FTashima-Tarsh\u002FDisha6.6\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fkd-esm\u002Freferences\u002Fworkflows.md",{"path":101,"resourceType":102,"resourceTypeDesc":103,"sizeBytes":104,"sourceHtmlUrl":105},"scripts\u002Fesm_embeddings.py","SCRIPT","脚本",4740,"https:\u002F\u002Fgithub.com\u002FTashima-Tarsh\u002FDisha6.6\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fkd-esm\u002Fscripts\u002Fesm_embeddings.py",[107,138,162,189,214,237],{"skillNo":108,"name":109,"nameZh":6,"description":110,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":111,"categories":112,"keywords":118,"publishedAt":38},"01M6QRD07D4TGNDK6VJ4EHF8BV","youtube-full","Use when the user needs YouTube transcripts, video search, channel browsing, playlist extraction, or content monitoring. Trigger phrases: 'get the transcript for', 'search YouTube for', 'what are the latest videos on', 'list this playlist', 'monitor this channel', or any request involving a YouTube URL, video ID, or @handle. Do NOT use for downloading video or audio files, YouTube engagement data (likes, comments), or private\u002Fage-restricted videos.","https:\u002F\u002Fgithub.com\u002FTashima-Tarsh\u002FDisha6.6\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fyoutube-full\u002FSKILL.md",[113],{"categoryNo":114,"categoryName":115,"taxonomyType":116,"taxonomyTypeDesc":117},"01JBSEEDCAT000000000000016","图像与多媒体","SCENARIO","使用场景",[119,120,121,122,123,124,125,126,127,128,129,130,131,132,133,134,20,135,136,137],"youtube full","youtube","full","transcript","search","monitoring","data","Video","Audio","list","files","do","get","age","trigger","URL","browsing","request","youtube transcripts",{"skillNo":139,"name":140,"nameZh":6,"description":141,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":142,"categories":143,"keywords":144,"publishedAt":38},"01M6QRCJ3112VYPA97YG7KGMC2","x-twitter-growth","X\u002FTwitter growth engine for building audience, crafting viral content, and analyzing engagement. Use when the user wants to grow on X\u002FTwitter, write tweets or threads, analyze their X profile, research competitors on X, plan a posting strategy, or optimize engagement. Complements social-content (generic multi-platform) with X-specific depth: algorithm mechanics, thread engineering, reply strategy, profile optimization, and competitive intelligence via web search.","https:\u002F\u002Fgithub.com\u002FTashima-Tarsh\u002FDisha6.6\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fx-twitter-growth\u002FSKILL.md",[],[145,146,147,148,123,149,150,151,152,153,154,155,156,157,26,158,159,160,161,20],"twitter","x twitter growth","growth","research","content","engineering","plan","web search","competitors","analyze","engine","write","strategy","platform","grow","posting","competitive",{"skillNo":163,"name":164,"nameZh":6,"description":165,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":166,"categories":167,"keywords":171,"publishedAt":38},"01M6QRBN3EHQNJY42RERP7A149","write-a-skill","Create new agent skills with proper structure, progressive disclosure, and bundled resources. Use when user wants to create, write, build, or author a new skill.","https:\u002F\u002Fgithub.com\u002FTashima-Tarsh\u002FDisha6.6\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fwrite-a-skill\u002FSKILL.md",[168],{"categoryNo":169,"categoryName":170,"taxonomyType":116,"taxonomyTypeDesc":117},"01JBSEEDCAT000000000000023","Skill与智能体开发",[172,156,173,174,175,176,177,178,26,179,180,51,181,182,183,184,185,186,187,188],"write a skill","github","create","https","productivity","matt pocock","direct","example","build","new skill","structure","new agent","resources","with","tree","progressive disclosure","author",{"skillNo":190,"name":191,"nameZh":6,"description":192,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":193,"categories":194,"keywords":195,"publishedAt":38},"01M6QRB6G8P33PNGAF9Y69EJ9T","wrap-up","Close out a launched Claude Managed Agent — recap every primitive the founder now owns, regenerate the single-file overview page, and suggest the next 1-2 upgrades. Use when the user says \"wrap up\", \"close this out\", \"what do I own now\", \"give me the summary\", \"recap the agent\", or when the orchestrator routes phase=wrap-up. primitives_inventory.py tables everything owned (agent, environment, session, memory, outcome, deployment); overview_page.py regenerates a self-contained .\u002Fmy-agent\u002Fagent-overview.html; upgrade_suggester.py ranks the next moves from recorded deferrals plus standing hardening steps. Companion to run-without-you; the last stop before phase=done.","https:\u002F\u002Fgithub.com\u002FTashima-Tarsh\u002FDisha6.6\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fwrap-up\u002FSKILL.md",[],[196,197,198,199,200,201,202,203,204,26,205,130,206,207,208,209,210,211,212,213],"wrap up","up","wrap","gemini","deployment","run","file","hardening","closeout","stop","upgrade","founder","close","companion","you","tables","directions","phase",{"skillNo":215,"name":216,"nameZh":6,"description":217,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":218,"categories":219,"keywords":221,"publishedAt":38},"01M6QRA5AYC1RBFZ39GTWGS3N8","workflow-builder","Design and write deterministic multi-agent workflow scripts (.js files in .claude\u002Fworkflows\u002F) for Claude Code's Workflow tool. Use when a user wants to build, create, author, scaffold, or run a custom Claude Code workflow, orchestrate sub-agents (fan-out, pipeline, loop, judge-panel), or automate a repeatable multi-step task across fresh-context agents.","https:\u002F\u002Fgithub.com\u002FTashima-Tarsh\u002FDisha6.6\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fworkflow-builder\u002FSKILL.md",[220],{"categoryNo":169,"categoryName":170,"taxonomyType":116,"taxonomyTypeDesc":117},[222,223,224,225,36,174,201,226,227,228,229,230,156,231,129,232,233,234,235,236],"workflow","workflow builder","builder","design","code","orchestrate","pipeline","loop","multi-agent","context","across","fresh","scripts","judge","claude code workflow",{"skillNo":238,"name":239,"nameZh":6,"description":240,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":241,"categories":242,"keywords":243,"publishedAt":38},"01M6QR7QMCZESKQA0ERW1E657M","weekly-review","Use when someone wants to run a weekly review, close open loops, audit stalled projects and commitments, get their system back to trusted, restart a lapsed review habit, or says \"\u002Fcs:weekly-review\". Walks David Allen's three-phase loop — GET CLEAR, GET CURRENT, GET CREATIVE — with deterministic scripts that inventory open loops, gate the checklist with named gaps, and score commitment health 0-100.","https:\u002F\u002Fgithub.com\u002FTashima-Tarsh\u002FDisha6.6\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fweekly-review\u002FSKILL.md",[],[244,245,246,247,248,249,250,251,252,229,253,254,255,256,257,258,131,259,260,213],"review","weekly review","weekly","deterministic scripts","reflect","feeds","inventory","open","team","projects","project-management","scanner","conversation","back","project","sprint","cadence"]