[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6QECJ8Y3XMF8E8G6JD6YZ8E":3,"skill-same:01M6QECJ8Y3XMF8E8G6JD6YZ8E":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":31,"publishedAt":51,"linkOnly":52,"markdownContent":53,"markdownBody":54,"frontMatter":55,"refName":61,"commitSha":62,"skillDirectory":63,"forksCount":14,"repositoryHtmlUrl":64,"repositoryHomepageUrl":65,"ownerHtmlUrl":66,"ownerBlogUrl":67,"ownerDescription":68,"updatedAt":69,"install":70,"resources":83},"01M6QECJ8Y3XMF8E8G6JD6YZ8E","esm",null,"Covers the EvolutionaryScale\u002FBiohub `esm` Python SDK: ESM3 generative protein design (sequence, structure and function tracks, chain-of-thought), ESMC embeddings, inverse folding, ESMFold2 structure prediction, and hosted Forge\u002FBiohub inference with ESM_API_KEY. Use when generating or completing proteins with ESM3, extracting ESMC embeddings for classification or similarity, folding a sequence or designing sequences from a structure, or choosing between esm3-open, esmc_300m and Forge-hosted models. Use when batching async Forge or Biohub API calls. Not for other protein models such as AlphaFold or generic Hugging Face transformers usage.","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\u002Fkal-esm\u002FSKILL.md",[18,23,26],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000010","后端与接口","SCENARIO","使用场景",{"categoryNo":24,"categoryName":25,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000015","视觉设计",{"categoryNo":27,"categoryName":28,"taxonomyType":29,"taxonomyTypeDesc":30},"01JBSEEDCAT000000000000032","设计师","PROFESSION","适用职业",[5,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50],"life","face","generative","sequence","inverse","extracting","chain-of-thought","chain","as","proteins","design","structure","hugging face","python sdk","open","api","Transformers","forge","embeddings","2026-11-10 10:12:30",false,"---\nname: esm\ndescription: 'Covers the EvolutionaryScale\u002FBiohub `esm` Python SDK: ESM3 generative protein design (sequence, structure and function tracks, chain-of-thought), ESMC embeddings, inverse folding, ESMFold2 structure prediction, and hosted Forge\u002FBiohub inference with ESM_API_KEY. Use when generating or completing proteins with ESM3, extracting ESMC embeddings for classification or similarity, folding a sequence or designing sequences from a structure, or choosing between esm3-open, esmc_300m and Forge-hosted models. Use when batching async Forge or Biohub API calls. Not for other protein models such as AlphaFold or generic Hugging Face transformers usage.'\nlicense: MIT\nmetadata:\n  version: '1.2'\n  category: life-sciences\n  maintainer: Kalaris Labs\n  tags: protein language models, protein design, structure prediction, protein embeddings\n---\n\n# ESM: Evolutionary Scale Modeling\n\n## Overview\n\nESM provides protein language models for understanding, generating, and designing proteins. Use this skill for current EvolutionaryScale\u002FBiohub workflows: ESM3 for generative design, ESMC for representation learning and embeddings, hosted Forge\u002FBiohub inference, and ESMFold2 all-atom structure prediction.\n\n## Core Capabilities\n\n### 1. Protein Sequence Generation with ESM3\n\nGenerate novel protein sequences with desired properties using multimodal generative modeling.\n\n**When to use:**\n- Designing proteins with specific functional properties\n- Completing partial protein sequences\n- Generating variants of existing proteins\n- Creating proteins with desired structural characteristics\n\n**Basic usage:**\n\n```python\nfrom esm.models.esm3 import ESM3\nfrom esm.sdk.api import ESM3InferenceClient, ESMProtein, GenerationConfig\n\n# Load local open weights after accepting the license on Hugging Face.\nmodel: ESM3InferenceClient = ESM3.from_pretrained(\"esm3-open\").to(\"cuda\")\n\n# Create protein prompt\nprotein = ESMProtein(sequence=\"MPRT___KEND\")  # '_' represents masked positions\n\n# Generate completion\nprotein = model.generate(protein, GenerationConfig(track=\"sequence\", num_steps=8))\nprint(protein.sequence)\n```\n\n**For remote\u002Fcloud usage via Forge API:**\n\n```python\nimport os\nimport esm\nfrom esm.sdk.api import ESMProtein, GenerationConfig\n\n# Same interface as local ESM3; token from ESM_API_KEY (see Authentication)\nmodel = esm.sdk.client(\"esm3-medium-2024-08\", token=os.environ[\"ESM_API_KEY\"])\n\n# Generate\nprotein = model.generate(protein, GenerationConfig(track=\"sequence\", num_steps=8))\n```\n\nSee `references\u002Fesm3-api.md` for detailed ESM3 model specifications, advanced generation configurations, and multimodal prompting examples.\n\n### 2. Structure Prediction and Inverse Folding\n\nUse ESM3's structure track for structure prediction from sequence or inverse folding (sequence design from structure).\n\n**Structure prediction:**\n\n```python\nfrom esm.sdk.api import ESM3InferenceClient, ESMProtein, GenerationConfig\n\n# Predict structure from sequence\nprotein = ESMProtein(sequence=\"MPRTKEINDAGLIVHSP...\")\nprotein_with_structure = model.generate(\n    protein,\n    GenerationConfig(track=\"structure\", num_steps=protein.sequence.count(\"_\"))\n)\n\n# Access predicted structure\ncoordinates = protein_with_structure.coordinates  # 3D coordinates\npdb_string = protein_with_structure.to_pdb()\n```\n\n**Inverse folding (sequence from structure):**\n\n```python\n# Design sequence for a target structure\nprotein_with_structure = ESMProtein.from_pdb(\"target_structure.pdb\")\nprotein_with_structure.sequence = None  # Remove sequence\n\n# Generate sequence that folds to this structure\ndesigned_protein = model.generate(\n    protein_with_structure,\n    GenerationConfig(track=\"sequence\", num_steps=50, temperature=0.7)\n)\n```\n\n### 3. Protein Embeddings with ESM C\n\nGenerate high-quality embeddings for downstream tasks like function prediction, classification, or similarity analysis.\n\n**When to use:**\n- Extracting protein representations for machine learning\n- Computing sequence similarities\n- Feature extraction for protein classification\n- Transfer learning for protein-related tasks\n\n**Basic usage:**\n\n```python\nfrom esm.models.esmc import ESMC\nfrom esm.sdk.api import ESMProtein, LogitsConfig\n\n# Load ESM C model\nmodel = ESMC.from_pretrained(\"esmc_300m\").to(\"cuda\")\n\n# Get embeddings\nprotein = ESMProtein(sequence=\"MPRTKEINDAGLIVHSP...\")\nprotein_tensor = model.encode(protein)\nlogits_output = model.logits(\n    protein_tensor,\n    LogitsConfig(sequence=True, return_embeddings=True),\n)\nembeddings = logits_output.embeddings\n```\n\n**Batch processing:**\n\n```python\n# Encode multiple proteins\nproteins = [\n    ESMProtein(sequence=\"MPRTKEIND...\"),\n    ESMProtein(sequence=\"AGLIVHSPQ...\"),\n    ESMProtein(sequence=\"KTEFLNDGR...\")\n]\n\nembeddings_list = [\n    model.logits(\n        model.encode(p),\n        LogitsConfig(sequence=True, return_embeddings=True),\n    ).embeddings\n    for p in proteins\n]\n```\n\nSee `references\u002Fesm-c-api.md` for ESM C model details, efficiency comparisons, and advanced embedding strategies.\n\n### 4. Function Conditioning and Annotation\n\nUse ESM3's function track to generate proteins with specific functional annotations or predict function from sequence.\n\n**Function-conditioned generation:**\n\n```python\nfrom esm.sdk.api import ESMProtein, FunctionAnnotation, GenerationConfig\n\n# Create protein with desired function\nprotein = ESMProtein(\n    sequence=\"_\" * 200,  # Generate 200 residue protein\n    function_annotations=[\n        FunctionAnnotation(label=\"fluorescent_protein\", start=50, end=150)\n    ]\n)\n\n# Generate sequence with specified function\nfunctional_protein = model.generate(\n    protein,\n    GenerationConfig(track=\"sequence\", num_steps=200)\n)\n```\n\n### 5. Chain-of-Thought Generation\n\nIteratively refine protein designs using ESM3's chain-of-thought generation approach.\n\n```python\nfrom esm.sdk.api import GenerationConfig\n\n# Multi-step refinement\nprotein = ESMProtein(sequence=\"MPRT\" + \"_\" * 100 + \"KEND\")\n\n# Step 1: Generate initial structure\nconfig = GenerationConfig(track=\"structure\", num_steps=50)\nprotein = model.generate(protein, config)\n\n# Step 2: Refine sequence based on structure\nconfig = GenerationConfig(track=\"sequence\", num_steps=50, temperature=0.5)\nprotein = model.generate(protein, config)\n\n# Step 3: Predict function\nconfig = GenerationConfig(track=\"function\", num_steps=20)\nprotein = model.generate(protein, config)\n```\n\n### 6. Batch Processing with Forge API\n\nProcess multiple proteins efficiently using Forge's async methods.\n\n```python\nimport os\nimport asyncio\nimport esm\nfrom esm.sdk.api import ESMProtein, GenerationConfig\n\nclient = esm.sdk.client(\"esm3-medium-2024-08\", token=os.environ[\"ESM_API_KEY\"])\n\n# Async batch processing\nasync def batch_generate(proteins_list):\n    tasks = [\n        client.async_generate(protein, GenerationConfig(track=\"sequence\"))\n        for protein in proteins_list\n    ]\n    return await asyncio.gather(*tasks)\n\n# Execute\nproteins = [ESMProtein(sequence=f\"MPRT{'_' * 50}KEND\") for _ in range(10)]\nresults = asyncio.run(batch_generate(proteins))\n```\n\nSee `references\u002Fforge-api.md` for detailed Forge API documentation, authentication, rate limits, and batch processing patterns.\n\n## Model Selection Guide\n\n**ESM3 Models (Generative):**\n- `esm3-open` (1.4B) - Open weights, local usage after accepting the Hugging Face license\n- `esm3-medium-2024-08` (7B) - Best balance of quality and speed (Forge only)\n- `esm3-large-2024-03` (98B) - Highest quality, slower (Forge only)\n\n**ESM C Models (Embeddings):**\n- `esmc_300m` \u002F `esmc-300m-2024-12` (30 layers) - Lightweight, fast inference (open weights, local)\n- `esmc_600m` \u002F `esmc-600m-2024-12` (36 layers) - Balanced performance (open weights, local)\n- `esmc-6b-2024-12` (80 layers) - Maximum quality (Forge API; local 6B weights require Forge or SageMaker)\n\nLocal `ESMC.from_pretrained()` examples use underscore aliases (`esmc_300m`, `esmc_600m`). Hosted API clients use dated model IDs such as `esmc-600m-2024-12`.\n\n**Selection criteria:**\n- **Local development\u002Ftesting:** Use `esm3-open` or `esmc_300m`\n- **Production quality:** Use `esm3-medium-2024-08` via Forge\n- **Maximum accuracy:** Use `esm3-large-2024-03` or `esmc-6b-2024-12` via Forge\n- **High throughput:** Use Forge or Biohub APIs with explicit async concurrency limits\n- **Cost optimization:** Use smaller models, implement caching strategies\n\n## Installation\n\nInstall from PyPI ([`esm` on PyPI](https:\u002F\u002Fpypi.org\u002Fproject\u002Fesm\u002F) by EvolutionaryScale). Current PyPI release: **3.2.3** (Oct 14, 2025). Requires **Python >=3.12,\u003C3.13**.\n\n**Basic installation:**\n\n```bash\nuv pip install \"esm==3.2.3\"\n```\n\n**With Flash Attention (recommended for faster inference on NVIDIA GPUs):**\n\n```bash\nuv pip install \"esm==3.2.3\"\nuv pip install flash-attn --no-build-isolation\n```\n\nThe Forge client ships with the `esm` package - no extra install for ESM3 or ESMC Forge inference.\n\n## Authentication\n\nForge API access requires an API key. Never hardcode tokens in scripts or commit them to version control.\n\n1. Check whether `ESM_API_KEY` is already set in the environment.\n2. If not, check a local `.env` for `ESM_API_KEY` only (do not load unrelated secrets).\n3. If still missing, create a key in the [Biohub developer console](https:\u002F\u002Fbiohub.ai\u002Fdeveloper-console\u002Fapi-keys) for Biohub APIs or [Forge](https:\u002F\u002Fforge.evolutionaryscale.ai) for legacy Forge-hosted ESM3\u002FESMC access.\n\n```python\nimport os\n\ntoken = os.environ[\"ESM_API_KEY\"]  # raises KeyError if unset\n```\n\n`esm.sdk.client()` reads `ESM_API_KEY` automatically when `token` is omitted. Keep endpoint URLs fixed to trusted hosts such as `https:\u002F\u002Fforge.evolutionaryscale.ai` or `https:\u002F\u002Fbiohub.ai`; do not take API hosts from untrusted user input.\n\n**Biohub platform:** EvolutionaryScale and Forge now surface current hosted models through [biohub.ai](https:\u002F\u002Fbiohub.ai). SDK class names may still reference \"Forge\". See `references\u002Fbiohub-platform.md` for ESMFold2 and Biohub-specific setup.\n\n## Common Workflows\n\nFor detailed examples and complete workflows, see `references\u002Fworkflows.md` which includes:\n- Novel GFP design with chain-of-thought\n- Protein variant generation and screening\n- Structure-based sequence optimization\n- Function prediction pipelines\n- Embedding-based clustering and analysis\n\n## References\n\nThis skill includes comprehensive reference documentation:\n\n- `references\u002Fesm3-api.md` - ESM3 model architecture, API reference, generation parameters, and multimodal prompting\n- `references\u002Fesm-c-api.md` - ESM C model details, embedding strategies, and performance optimization\n- `references\u002Fforge-api.md` - Forge platform documentation, authentication, batch processing, and deployment\n- `references\u002Fbiohub-platform.md` - Biohub API migration, ESMFold2 structure prediction, and developer-console auth\n- `references\u002Fworkflows.md` - Complete examples and common workflow patterns\n\nThese references contain detailed API specifications, parameter descriptions, and advanced usage patterns. Load them as needed for specific tasks.\n\n## Best Practices\n\n**For generation tasks:**\n- Start with smaller models for prototyping (`esm3-open`)\n- Use temperature parameter to control diversity (0.0 = deterministic, 1.0 = diverse)\n- Implement iterative refinement with chain-of-thought for complex designs\n- Validate generated sequences with structure prediction or wet-lab experiments\n\n**For embedding tasks:**\n- Batch process sequences when possible for efficiency\n- Cache embeddings for repeated analyses\n- Normalize embeddings when computing similarities\n- Use appropriate model size based on downstream task requirements\n\n**For production deployment:**\n- Use Forge API for scalability and latest models\n- Implement error handling and retry logic for API calls\n- Monitor token usage and implement rate limiting\n- Consider AWS SageMaker deployment for dedicated infrastructure\n\n## Resources and Documentation\n\n- **GitHub Repository:** https:\u002F\u002Fgithub.com\u002FBiohub\u002Fesm (current ESMC\u002FESMFold2\u002FBiohub docs; ESM3 docs remain linked from the repository)\n- **Forge Platform:** https:\u002F\u002Fforge.evolutionaryscale.ai\n- **Biohub Platform:** https:\u002F\u002Fbiohub.ai\n- **Scientific Paper:** Hayes et al., Science (2025) - https:\u002F\u002Fwww.science.org\u002Fdoi\u002F10.1126\u002Fscience.ads0018\n- **Blog Posts:**\n  - ESM3 Release: https:\u002F\u002Fwww.evolutionaryscale.ai\u002Fblog\u002Fesm3-release\n  - ESM C Launch: https:\u002F\u002Fwww.evolutionaryscale.ai\u002Fblog\u002Fesm-cambrian\n- **Community:** Slack community at https:\u002F\u002Fbit.ly\u002F3FKwcWd\n- **Model Weights:** Hugging Face EvolutionaryScale and Biohub organizations\n\n## Responsible Use\n\nESM is designed for beneficial applications in protein engineering, drug discovery, and scientific research. Follow the Responsible Biodesign Framework (https:\u002F\u002Fresponsiblebiodesign.ai\u002F) and Biohub Acceptable Use Policy (https:\u002F\u002Fbiohub.org\u002Facceptable-use-policy\u002F) when designing novel proteins. Consider biosafety and ethical implications of protein designs before experimental validation.\n\n## Agent operating procedure\n\n1. **Check the environment.** Confirm tool versions, the reference genome\u002Fannotation build and the input formats (FASTQ, BAM, VCF, h5ad).\n2. **Pin down the inputs.** Confirm formats, identifiers and parameters from the data or the user. Ask rather than guess any value that changes the result.\n3. **Run a small version first.** Run the pipeline on a small subset (one sample, one chromosome, a few thousand cells) first.\n4. **Execute the full task** using the instructions and references above.\n5. **Validate the result.** Check QC metrics, sample identities, genome build consistency and batch effects before interpreting results.\n6. **Report.** State what was run (versions, commands, parameters), what was checked, and what is still uncertain.\n\n| If this happens | Do this |\n|---|---|\n| Genome builds or identifiers do not match between inputs | Stop and harmonize (liftover, ID mapping) before continuing. |\n| A function, flag or endpoint in these instructions is missing in the installed version | Check the installed version's own documentation (`help()`, `--help`, official docs), adapt, and tell the user. Never invent an API. |\n| A required input, identifier or parameter is ambiguous | Ask the user, or state the assumption explicitly before running. |\n\n**Integrity rules**\n\n- Never fabricate results, parameters, identifiers, citations or statistics. If something cannot be run or verified, say so plainly.\n- Do not interpret biological significance beyond what the statistics support; report multiple-testing correction.\n- Treat version-specific details here as possibly outdated: confirm them against the official documentation for the installed version.\n- Ask before actions that cost money, consume shared GPUs or cloud quota, touch personal or patient data, or cannot be undone.\n\n## Related skills\n\n- `sentence-transformers`: Generates sentence, text, and image embeddings locally with the Python sentence-transformers (SBERT) library, using pre-trained Hugging Fac…\n- `adaptyv`: How to use the Adaptyv Bio Foundry API and Python SDK for protein experiment design, submission, and results retrieval.\n- `tamarind`: Access a collection of open-source molecular design and structural biology tools on the Tamarind Bio platform, via its REST API or MCP serv…\n","# ESM: Evolutionary Scale Modeling\n\n## Overview\n\nESM provides protein language models for understanding, generating, and designing proteins. Use this skill for current EvolutionaryScale\u002FBiohub workflows: ESM3 for generative design, ESMC for representation learning and embeddings, hosted Forge\u002FBiohub inference, and ESMFold2 all-atom structure prediction.\n\n## Core Capabilities\n\n### 1. Protein Sequence Generation with ESM3\n\nGenerate novel protein sequences with desired properties using multimodal generative modeling.\n\n**When to use:**\n- Designing proteins with specific functional properties\n- Completing partial protein sequences\n- Generating variants of existing proteins\n- Creating proteins with desired structural characteristics\n\n**Basic usage:**\n\n```python\nfrom esm.models.esm3 import ESM3\nfrom esm.sdk.api import ESM3InferenceClient, ESMProtein, GenerationConfig\n\n# Load local open weights after accepting the license on Hugging Face.\nmodel: ESM3InferenceClient = ESM3.from_pretrained(\"esm3-open\").to(\"cuda\")\n\n# Create protein prompt\nprotein = ESMProtein(sequence=\"MPRT___KEND\")  # '_' represents masked positions\n\n# Generate completion\nprotein = model.generate(protein, GenerationConfig(track=\"sequence\", num_steps=8))\nprint(protein.sequence)\n```\n\n**For remote\u002Fcloud usage via Forge API:**\n\n```python\nimport os\nimport esm\nfrom esm.sdk.api import ESMProtein, GenerationConfig\n\n# Same interface as local ESM3; token from ESM_API_KEY (see Authentication)\nmodel = esm.sdk.client(\"esm3-medium-2024-08\", token=os.environ[\"ESM_API_KEY\"])\n\n# Generate\nprotein = model.generate(protein, GenerationConfig(track=\"sequence\", num_steps=8))\n```\n\nSee `references\u002Fesm3-api.md` for detailed ESM3 model specifications, advanced generation configurations, and multimodal prompting examples.\n\n### 2. Structure Prediction and Inverse Folding\n\nUse ESM3's structure track for structure prediction from sequence or inverse folding (sequence design from structure).\n\n**Structure prediction:**\n\n```python\nfrom esm.sdk.api import ESM3InferenceClient, ESMProtein, GenerationConfig\n\n# Predict structure from sequence\nprotein = ESMProtein(sequence=\"MPRTKEINDAGLIVHSP...\")\nprotein_with_structure = model.generate(\n    protein,\n    GenerationConfig(track=\"structure\", num_steps=protein.sequence.count(\"_\"))\n)\n\n# Access predicted structure\ncoordinates = protein_with_structure.coordinates  # 3D coordinates\npdb_string = protein_with_structure.to_pdb()\n```\n\n**Inverse folding (sequence from structure):**\n\n```python\n# Design sequence for a target structure\nprotein_with_structure = ESMProtein.from_pdb(\"target_structure.pdb\")\nprotein_with_structure.sequence = None  # Remove sequence\n\n# Generate sequence that folds to this structure\ndesigned_protein = model.generate(\n    protein_with_structure,\n    GenerationConfig(track=\"sequence\", num_steps=50, temperature=0.7)\n)\n```\n\n### 3. Protein Embeddings with ESM C\n\nGenerate high-quality embeddings for downstream tasks like function prediction, classification, or similarity analysis.\n\n**When to use:**\n- Extracting protein representations for machine learning\n- Computing sequence similarities\n- Feature extraction for protein classification\n- Transfer learning for protein-related tasks\n\n**Basic usage:**\n\n```python\nfrom esm.models.esmc import ESMC\nfrom esm.sdk.api import ESMProtein, LogitsConfig\n\n# Load ESM C model\nmodel = ESMC.from_pretrained(\"esmc_300m\").to(\"cuda\")\n\n# Get embeddings\nprotein = ESMProtein(sequence=\"MPRTKEINDAGLIVHSP...\")\nprotein_tensor = model.encode(protein)\nlogits_output = model.logits(\n    protein_tensor,\n    LogitsConfig(sequence=True, return_embeddings=True),\n)\nembeddings = logits_output.embeddings\n```\n\n**Batch processing:**\n\n```python\n# Encode multiple proteins\nproteins = [\n    ESMProtein(sequence=\"MPRTKEIND...\"),\n    ESMProtein(sequence=\"AGLIVHSPQ...\"),\n    ESMProtein(sequence=\"KTEFLNDGR...\")\n]\n\nembeddings_list = [\n    model.logits(\n        model.encode(p),\n        LogitsConfig(sequence=True, return_embeddings=True),\n    ).embeddings\n    for p in proteins\n]\n```\n\nSee `references\u002Fesm-c-api.md` for ESM C model details, efficiency comparisons, and advanced embedding strategies.\n\n### 4. Function Conditioning and Annotation\n\nUse ESM3's function track to generate proteins with specific functional annotations or predict function from sequence.\n\n**Function-conditioned generation:**\n\n```python\nfrom esm.sdk.api import ESMProtein, FunctionAnnotation, GenerationConfig\n\n# Create protein with desired function\nprotein = ESMProtein(\n    sequence=\"_\" * 200,  # Generate 200 residue protein\n    function_annotations=[\n        FunctionAnnotation(label=\"fluorescent_protein\", start=50, end=150)\n    ]\n)\n\n# Generate sequence with specified function\nfunctional_protein = model.generate(\n    protein,\n    GenerationConfig(track=\"sequence\", num_steps=200)\n)\n```\n\n### 5. Chain-of-Thought Generation\n\nIteratively refine protein designs using ESM3's chain-of-thought generation approach.\n\n```python\nfrom esm.sdk.api import GenerationConfig\n\n# Multi-step refinement\nprotein = ESMProtein(sequence=\"MPRT\" + \"_\" * 100 + \"KEND\")\n\n# Step 1: Generate initial structure\nconfig = GenerationConfig(track=\"structure\", num_steps=50)\nprotein = model.generate(protein, config)\n\n# Step 2: Refine sequence based on structure\nconfig = GenerationConfig(track=\"sequence\", num_steps=50, temperature=0.5)\nprotein = model.generate(protein, config)\n\n# Step 3: Predict function\nconfig = GenerationConfig(track=\"function\", num_steps=20)\nprotein = model.generate(protein, config)\n```\n\n### 6. Batch Processing with Forge API\n\nProcess multiple proteins efficiently using Forge's async methods.\n\n```python\nimport os\nimport asyncio\nimport esm\nfrom esm.sdk.api import ESMProtein, GenerationConfig\n\nclient = esm.sdk.client(\"esm3-medium-2024-08\", token=os.environ[\"ESM_API_KEY\"])\n\n# Async batch processing\nasync def batch_generate(proteins_list):\n    tasks = [\n        client.async_generate(protein, GenerationConfig(track=\"sequence\"))\n        for protein in proteins_list\n    ]\n    return await asyncio.gather(*tasks)\n\n# Execute\nproteins = [ESMProtein(sequence=f\"MPRT{'_' * 50}KEND\") for _ in range(10)]\nresults = asyncio.run(batch_generate(proteins))\n```\n\nSee `references\u002Fforge-api.md` for detailed Forge API documentation, authentication, rate limits, and batch processing patterns.\n\n## Model Selection Guide\n\n**ESM3 Models (Generative):**\n- `esm3-open` (1.4B) - Open weights, local usage after accepting the Hugging Face license\n- `esm3-medium-2024-08` (7B) - Best balance of quality and speed (Forge only)\n- `esm3-large-2024-03` (98B) - Highest quality, slower (Forge only)\n\n**ESM C Models (Embeddings):**\n- `esmc_300m` \u002F `esmc-300m-2024-12` (30 layers) - Lightweight, fast inference (open weights, local)\n- `esmc_600m` \u002F `esmc-600m-2024-12` (36 layers) - Balanced performance (open weights, local)\n- `esmc-6b-2024-12` (80 layers) - Maximum quality (Forge API; local 6B weights require Forge or SageMaker)\n\nLocal `ESMC.from_pretrained()` examples use underscore aliases (`esmc_300m`, `esmc_600m`). Hosted API clients use dated model IDs such as `esmc-600m-2024-12`.\n\n**Selection criteria:**\n- **Local development\u002Ftesting:** Use `esm3-open` or `esmc_300m`\n- **Production quality:** Use `esm3-medium-2024-08` via Forge\n- **Maximum accuracy:** Use `esm3-large-2024-03` or `esmc-6b-2024-12` via Forge\n- **High throughput:** Use Forge or Biohub APIs with explicit async concurrency limits\n- **Cost optimization:** Use smaller models, implement caching strategies\n\n## Installation\n\nInstall from PyPI ([`esm` on PyPI](https:\u002F\u002Fpypi.org\u002Fproject\u002Fesm\u002F) by EvolutionaryScale). Current PyPI release: **3.2.3** (Oct 14, 2025). Requires **Python >=3.12,\u003C3.13**.\n\n**Basic installation:**\n\n```bash\nuv pip install \"esm==3.2.3\"\n```\n\n**With Flash Attention (recommended for faster inference on NVIDIA GPUs):**\n\n```bash\nuv pip install \"esm==3.2.3\"\nuv pip install flash-attn --no-build-isolation\n```\n\nThe Forge client ships with the `esm` package - no extra install for ESM3 or ESMC Forge inference.\n\n## Authentication\n\nForge API access requires an API key. Never hardcode tokens in scripts or commit them to version control.\n\n1. Check whether `ESM_API_KEY` is already set in the environment.\n2. If not, check a local `.env` for `ESM_API_KEY` only (do not load unrelated secrets).\n3. If still missing, create a key in the [Biohub developer console](https:\u002F\u002Fbiohub.ai\u002Fdeveloper-console\u002Fapi-keys) for Biohub APIs or [Forge](https:\u002F\u002Fforge.evolutionaryscale.ai) for legacy Forge-hosted ESM3\u002FESMC access.\n\n```python\nimport os\n\ntoken = os.environ[\"ESM_API_KEY\"]  # raises KeyError if unset\n```\n\n`esm.sdk.client()` reads `ESM_API_KEY` automatically when `token` is omitted. Keep endpoint URLs fixed to trusted hosts such as `https:\u002F\u002Fforge.evolutionaryscale.ai` or `https:\u002F\u002Fbiohub.ai`; do not take API hosts from untrusted user input.\n\n**Biohub platform:** EvolutionaryScale and Forge now surface current hosted models through [biohub.ai](https:\u002F\u002Fbiohub.ai). SDK class names may still reference \"Forge\". See `references\u002Fbiohub-platform.md` for ESMFold2 and Biohub-specific setup.\n\n## Common Workflows\n\nFor detailed examples and complete workflows, see `references\u002Fworkflows.md` which includes:\n- Novel GFP design with chain-of-thought\n- Protein variant generation and screening\n- Structure-based sequence optimization\n- Function prediction pipelines\n- Embedding-based clustering and analysis\n\n## References\n\nThis skill includes comprehensive reference documentation:\n\n- `references\u002Fesm3-api.md` - ESM3 model architecture, API reference, generation parameters, and multimodal prompting\n- `references\u002Fesm-c-api.md` - ESM C model details, embedding strategies, and performance optimization\n- `references\u002Fforge-api.md` - Forge platform documentation, authentication, batch processing, and deployment\n- `references\u002Fbiohub-platform.md` - Biohub API migration, ESMFold2 structure prediction, and developer-console auth\n- `references\u002Fworkflows.md` - Complete examples and common workflow patterns\n\nThese references contain detailed API specifications, parameter descriptions, and advanced usage patterns. Load them as needed for specific tasks.\n\n## Best Practices\n\n**For generation tasks:**\n- Start with smaller models for prototyping (`esm3-open`)\n- Use temperature parameter to control diversity (0.0 = deterministic, 1.0 = diverse)\n- Implement iterative refinement with chain-of-thought for complex designs\n- Validate generated sequences with structure prediction or wet-lab experiments\n\n**For embedding tasks:**\n- Batch process sequences when possible for efficiency\n- Cache embeddings for repeated analyses\n- Normalize embeddings when computing similarities\n- Use appropriate model size based on downstream task requirements\n\n**For production deployment:**\n- Use Forge API for scalability and latest models\n- Implement error handling and retry logic for API calls\n- Monitor token usage and implement rate limiting\n- Consider AWS SageMaker deployment for dedicated infrastructure\n\n## Resources and Documentation\n\n- **GitHub Repository:** https:\u002F\u002Fgithub.com\u002FBiohub\u002Fesm (current ESMC\u002FESMFold2\u002FBiohub docs; ESM3 docs remain linked from the repository)\n- **Forge Platform:** https:\u002F\u002Fforge.evolutionaryscale.ai\n- **Biohub Platform:** https:\u002F\u002Fbiohub.ai\n- **Scientific Paper:** Hayes et al., Science (2025) - https:\u002F\u002Fwww.science.org\u002Fdoi\u002F10.1126\u002Fscience.ads0018\n- **Blog Posts:**\n  - ESM3 Release: https:\u002F\u002Fwww.evolutionaryscale.ai\u002Fblog\u002Fesm3-release\n  - ESM C Launch: https:\u002F\u002Fwww.evolutionaryscale.ai\u002Fblog\u002Fesm-cambrian\n- **Community:** Slack community at https:\u002F\u002Fbit.ly\u002F3FKwcWd\n- **Model Weights:** Hugging Face EvolutionaryScale and Biohub organizations\n\n## Responsible Use\n\nESM is designed for beneficial applications in protein engineering, drug discovery, and scientific research. Follow the Responsible Biodesign Framework (https:\u002F\u002Fresponsiblebiodesign.ai\u002F) and Biohub Acceptable Use Policy (https:\u002F\u002Fbiohub.org\u002Facceptable-use-policy\u002F) when designing novel proteins. Consider biosafety and ethical implications of protein designs before experimental validation.\n\n## Agent operating procedure\n\n1. **Check the environment.** Confirm tool versions, the reference genome\u002Fannotation build and the input formats (FASTQ, BAM, VCF, h5ad).\n2. **Pin down the inputs.** Confirm formats, identifiers and parameters from the data or the user. Ask rather than guess any value that changes the result.\n3. **Run a small version first.** Run the pipeline on a small subset (one sample, one chromosome, a few thousand cells) first.\n4. **Execute the full task** using the instructions and references above.\n5. **Validate the result.** Check QC metrics, sample identities, genome build consistency and batch effects before interpreting results.\n6. **Report.** State what was run (versions, commands, parameters), what was checked, and what is still uncertain.\n\n| If this happens | Do this |\n|---|---|\n| Genome builds or identifiers do not match between inputs | Stop and harmonize (liftover, ID mapping) before continuing. |\n| A function, flag or endpoint in these instructions is missing in the installed version | Check the installed version's own documentation (`help()`, `--help`, official docs), adapt, and tell the user. Never invent an API. |\n| A required input, identifier or parameter is ambiguous | Ask the user, or state the assumption explicitly before running. |\n\n**Integrity rules**\n\n- Never fabricate results, parameters, identifiers, citations or statistics. If something cannot be run or verified, say so plainly.\n- Do not interpret biological significance beyond what the statistics support; report multiple-testing correction.\n- Treat version-specific details here as possibly outdated: confirm them against the official documentation for the installed version.\n- Ask before actions that cost money, consume shared GPUs or cloud quota, touch personal or patient data, or cannot be undone.\n\n## Related skills\n\n- `sentence-transformers`: Generates sentence, text, and image embeddings locally with the Python sentence-transformers (SBERT) library, using pre-trained Hugging Fac…\n- `adaptyv`: How to use the Adaptyv Bio Foundry API and Python SDK for protein experiment design, submission, and results retrieval.\n- `tamarind`: Access a collection of open-source molecular design and structural biology tools on the Tamarind Bio platform, via its REST API or MCP serv…\n",{"name":5,"license":15,"metadata":56,"description":7},{"tags":57,"version":58,"category":59,"maintainer":60},"protein language models, protein design, structure prediction, protein embeddings","1.2","life-sciences","Kalaris Labs","main","0da4485785ee599388900be50db1635489d53056",".agents\u002Fskills\u002Fkal-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":71,"prompt":81,"zipAvailable":82},[72,75,78],{"tool":73,"command":74},"npx","npx skills add Tashima-Tarsh\u002FDisha6.6 --skill esm",{"tool":76,"command":77},"bunx","bunx skills add Tashima-Tarsh\u002FDisha6.6 --skill esm",{"tool":79,"command":80},"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\u002Fkal-esm\u002FSKILL.md",true,[84,90,94,98,102],{"path":85,"resourceType":86,"resourceTypeDesc":87,"sizeBytes":88,"sourceHtmlUrl":89},"references\u002Fbiohub-platform.md","REFERENCE","参考文档",4579,"https:\u002F\u002Fgithub.com\u002FTashima-Tarsh\u002FDisha6.6\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fkal-esm\u002Freferences\u002Fbiohub-platform.md",{"path":91,"resourceType":86,"resourceTypeDesc":87,"sizeBytes":92,"sourceHtmlUrl":93},"references\u002Fesm-c-api.md",16860,"https:\u002F\u002Fgithub.com\u002FTashima-Tarsh\u002FDisha6.6\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fkal-esm\u002Freferences\u002Fesm-c-api.md",{"path":95,"resourceType":86,"resourceTypeDesc":87,"sizeBytes":96,"sourceHtmlUrl":97},"references\u002Fesm3-api.md",12101,"https:\u002F\u002Fgithub.com\u002FTashima-Tarsh\u002FDisha6.6\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fkal-esm\u002Freferences\u002Fesm3-api.md",{"path":99,"resourceType":86,"resourceTypeDesc":87,"sizeBytes":100,"sourceHtmlUrl":101},"references\u002Fforge-api.md",18888,"https:\u002F\u002Fgithub.com\u002FTashima-Tarsh\u002FDisha6.6\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fkal-esm\u002Freferences\u002Fforge-api.md",{"path":103,"resourceType":86,"resourceTypeDesc":87,"sizeBytes":104,"sourceHtmlUrl":105},"references\u002Fworkflows.md",20643,"https:\u002F\u002Fgithub.com\u002FTashima-Tarsh\u002FDisha6.6\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fkal-esm\u002Freferences\u002Fworkflows.md",[107,137,162,188,213,236],{"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":116,"publishedAt":51},"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":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000016","图像与多媒体",[117,118,119,120,121,122,123,124,125,126,127,128,129,130,131,132,133,134,135,136],"youtube full","youtube","full","transcript","search","monitoring","data","Video","Audio","list","files","do","get","age","trigger","URL","for","browsing","request","youtube transcripts",{"skillNo":138,"name":139,"nameZh":6,"description":140,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":141,"categories":142,"keywords":143,"publishedAt":51},"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",[],[144,145,146,147,121,148,149,150,151,152,153,154,155,156,157,158,159,160,161,133],"twitter","x twitter growth","growth","research","content","engineering","plan","web search","competitors","analyze","engine","write","strategy","use","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":51},"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":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000023","Skill与智能体开发",[172,155,173,174,175,176,177,178,157,179,180,61,181,43,182,183,184,185,186,187],"write a skill","github","create","https","productivity","matt pocock","direct","example","build","new skill","new agent","resources","with","tree","progressive disclosure","author",{"skillNo":189,"name":190,"nameZh":6,"description":191,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":192,"categories":193,"keywords":194,"publishedAt":51},"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",[],[195,196,197,198,199,200,201,202,203,157,204,128,205,206,207,208,209,210,211,212],"wrap up","up","wrap","gemini","deployment","run","file","hardening","closeout","stop","upgrade","founder","close","companion","you","tables","directions","phase",{"skillNo":214,"name":215,"nameZh":6,"description":216,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":217,"categories":218,"keywords":220,"publishedAt":51},"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",[219],{"categoryNo":169,"categoryName":170,"taxonomyType":21,"taxonomyTypeDesc":22},[221,222,223,42,224,174,200,225,226,227,228,229,155,230,127,231,232,233,234,235],"workflow","workflow builder","builder","workflows","code","orchestrate","pipeline","loop","multi-agent","context","across","fresh","scripts","judge","claude code workflow",{"skillNo":237,"name":238,"nameZh":6,"description":239,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":240,"categories":241,"keywords":242,"publishedAt":51},"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",[],[243,244,245,246,247,248,249,46,250,228,251,252,253,254,255,256,129,257,258,212],"review","weekly review","weekly","deterministic scripts","reflect","feeds","inventory","team","projects","project-management","scanner","conversation","back","project","sprint","cadence"]