Agent skill

Boltz2 Binding Affinity

by InternScience in InternScience/scp

Predict protein-ligand binding affinity using Boltz-2 model to assess molecular interactions and binding probability for drug discovery.

MITAuto-check: warningsResearch & Science

Install Boltz2 Binding Affinity

The automated check flagged lines worth reading first. See the safety section below.

skills CLI
$ npx skills add InternScience/scp --skill boltz2-binding-affinity -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install InternScience/scp boltz2-binding-affinity --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/InternScience/scp.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/boltz2-binding-affinity .claude/skills/boltz2-binding-affinity && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
boltz2-binding-affinity
GitHub stars
169
Used in
1 other repo
Token cost
~1.4k tokens
SKILL.md length
266 words
Files
1
Skills in repo
73
Repo updated
First seen
Licence
MIT

At a glance

Predict protein-ligand binding affinity using Boltz-2 model to assess molecular interactions and binding probability for drug discovery.

  • Works in 2 steps: MCP Server Definition → Boltz-2 Binding Affinity Workflow
  • Tasks that involve Drug discovery and cheminformatics
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md

What it does

Boltz2 Binding Affinity is an agent skill from InternScience/scp. Predict protein-ligand binding affinity using Boltz-2 model to assess molecular interactions and binding probability for drug discovery.

Its SKILL.md is about 1.4k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Research & Science, covering Drug discovery and cheminformatics. The licence is MIT.

When your agent uses it

  • Tasks that involve Drug discovery and cheminformatics

Example prompts

  • “/boltz2-binding-affinity”

Requirements

  • Python 3

Workflow steps

2 steps, taken from the step headings in SKILL.md.

  1. MCP Server Definition
  2. Boltz-2 Binding Affinity Workflow

What it can do on your machine

Read from SKILL.md and the folder at commit cea5398. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md (its code samples are python).

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Boltz2 Binding Affinity loads about 1.4k tokens when it runs. Until then it costs about 40 tokens; SKILL.md has 266 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~40
When it runs · the whole SKILL.md, loaded when a task matches
~1.4k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check: warnings

The automated check found patterns that need a careful read before installing.

  • WarningContains a long base64 blobSKILL.md:101
    sequence = 'PIVQNLQGQMVHQCISPRTLNAWVKVVEEKAFSPEVIPMFSALSCGATPQDLNTMLNTVGGHQAAMQMLKETINEEAAEWDRLHPVHAGPIAPGQMREPRGSDIAGTT

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from InternScience/scp at commit cea5398, republished under its MIT licence (© InternScience). 266 words, ~1,443 tokens.

Download SKILL.mdSave it as .claude/skills/boltz2-binding-affinity/SKILL.md (or your agent's skills folder).
name
boltz2-binding-affinity
description
Predict protein-ligand binding affinity using Boltz-2 model to assess molecular interactions and binding probability for drug discovery.
license
MIT license
metadata.skill-author
PJLab

Boltz-2 Protein-Ligand Binding Affinity Prediction

Usage

1. MCP Server Definition
python
import asyncio
import json
from mcp.client.streamable_http import streamablehttp_client
from mcp import ClientSession

class DrugSDAClient:
    """DrugSDA-Model MCP Client"""

    def __init__(self, server_url: str, api_key: str):
        self.server_url = server_url
        self.api_key = api_key
        self.session = None

    async def connect(self):
        """Establish connection and initialize session"""
        print(f"server url: {self.server_url}")
        try:
            self.transport = streamablehttp_client(
                url=self.server_url,
                headers={"SCP-HUB-API-KEY": self.api_key}
            )
            self.read, self.write, self.get_session_id = await self.transport.__aenter__()

            self.session_ctx = ClientSession(self.read, self.write)
            self.session = await self.session_ctx.__aenter__()

            await self.session.initialize()
            session_id = self.get_session_id()

            print(f"✓ connect success")
            return True

        except Exception as e:
            print(f"✗ connect failure: {e}")
            import traceback
            traceback.print_exc()
            return False

    async def disconnect(self):
        """Disconnect from server"""
        try:
            if self.session:
                await self.session_ctx.__aexit__(None, None, None)
            if hasattr(self, 'transport'):
                await self.transport.__aexit__(None, None, None)
            print("✓ already disconnect")
        except Exception as e:
            print(f"✗ disconnect error: {e}")

    def parse_result(self, result):
        """Parse MCP tool call result"""
        try:
            if hasattr(result, 'content') and result.content:
                content = result.content[0]
                if hasattr(content, 'text'):
                    return json.loads(content.text)
            return str(result)
        except Exception as e:
            return {"error": f"parse error: {e}", "raw": str(result)}
2. Boltz-2 Binding Affinity Workflow

This workflow predicts protein-ligand binding affinity using the Boltz-2 deep learning model, providing affinity probabilities and 3D complex structures.

Workflow Steps:

  1. Prepare Input - Define protein sequence and SMILES list for ligands
  2. Run Boltz-2 Prediction - Calculate binding affinity probability for each ligand
  3. Analyze Results - Extract affinity scores and structure files

Implementation:

python
## Initialize client
client = DrugSDAClient(
    "https://scp.intern-ai.org.cn/api/v1/mcp/3/DrugSDA-Model",
    "<your-api-key>"
)

if not await client.connect():
    print("connection failed")
    exit()

## Input: Protein sequence and ligand SMILES
sequence = 'PIVQNLQGQMVHQCISPRTLNAWVKVVEEKAFSPEVIPMFSALSCGATPQDLNTMLNTVGGHQAAMQMLKETINEEAAEWDRLHPVHAGPIAPGQMREPRGSDIAGTTSTLQEQIGWMTHNPPIPVGEIYKRWIILGLNKIVRMYSPTSILDIRQGPKEPFRDYVDRFYKTLRAEQASQEVKNAATETLLVQNANPDCKTILKALGPGATLEEMMTACQG'
protein = [{'chain': 'A', 'sequence': sequence}]
smiles_list = ['N[C@@H](Cc1ccc(O)cc1)C(=O)O', "CC(C)C1=CC=CC=C1"]

## Execute Boltz-2 binding affinity prediction
result = await client.session.call_tool(
    "boltz_binding_affinity",
    arguments={
        "protein": protein,
        "smiles_list": smiles_list
    }
)

result_data = client.parse_result(result)
boltz_res = result_data["boltz_res"]

## Display results
for i, item in enumerate(boltz_res, 1):
    print(f"{i}. SMILES: {item['smiles']}")
    print(f"   Affinity Probability: {item['affinity_probability']:.4f}")
    print(f"   Structure File: {item['cif_file']}\n")

await client.disconnect()
Tool Descriptions

DrugSDA-Model Server:

  • boltz_binding_affinity: Predict protein-ligand binding affinity using Boltz-2
    • Args:
      • protein (list): List of protein chains with sequence information
        • Each chain: {'chain': str, 'sequence': str}
      • smiles_list (list): List of ligand SMILES strings
    • Returns:
      • boltz_res (list): List of binding predictions
        • smiles (str): Ligand SMILES string
        • affinity_probability (float): Binding affinity probability (0-1)
        • cif_file (str): Path to predicted complex structure
Input/Output

Input:

  • protein: List of protein chains
    • chain: Chain identifier (e.g., 'A', 'B')
    • sequence: Amino acid sequence in single-letter code
  • smiles_list: List of SMILES strings for ligand molecules

Output:

  • List of binding predictions, each containing:
    • smiles: Ligand SMILES string
    • affinity_probability: Binding probability (0-1, higher is better)
    • cif_file: Path to predicted protein-ligand complex structure in CIF format
Affinity Interpretation
  • Probability > 0.5: Strong binding likelihood
  • Probability 0.3-0.5: Moderate binding potential
  • Probability < 0.3: Weak or no binding expected
Use Cases
  • Virtual screening of compound libraries
  • Lead optimization in drug discovery
  • Protein-ligand binding mode prediction
  • Structure-based drug design
  • Comparative binding analysis across ligands
Performance Notes
  • Execution time: 30-120 seconds per ligand depending on protein size
  • Protein length: Best for proteins <1000 amino acids
  • Multiple ligands: Processes sequentially, allow sufficient time
  • Structure output: CIF files can be visualized in PyMOL, ChimeraX, or similar tools

© InternScience, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in skills/boltz2-binding-affinity of InternScience/scp.

Open the folder on GitHubat commit cea5398

Used in 2 other repositories

We found 2 copies of this SKILL.md (exact, near-identical or edited) in other folders, from 1 other GitHub owner. This page covers the copy in InternScience/scp, which our catalogue first saw on October 7, 2026.

Compare with similar skills

Boltz2 Binding Affinity next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Boltz2 Binding Affinity compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Boltz2 Binding Affinity this skillInternScience/scp1691 repos~1.4kAutomated safety check: WarnMIT
MolecodeAtomFlow-AI/MoleCode305—~1.9kAutomated safety check: PassMIT
Drug DiscoveryTommy-yw/RunbookHermes5461 repos~2.3kAutomated safety check: PassMIT
DiffDock Molecular DockingK-Dense-AI/scientific-agent-skills48k1 repos~3kAutomated safety check: NotesMIT
Biomedical Analysis Dispatchxjtulyc/MedgeClaw6171 repos~2kAutomated safety check: PassNone
Edu Chem Reactionwy51ai/edulab1.4k—~1.2kAutomated safety check: PassApache-2.0

Similar skills

  • Molecode

    AtomFlow-AI/MoleCode

    A skill your agent uses for deterministic molecule understanding, graph-level editing, generation, and validation with MoleCode — an explicit Mermaid graph in which every atom and bond is a typed…

    305 GitHub stars~1.9k tokensUpdated 4 mo ago
    Research & ScienceAuto-check passed
  • Drug Discovery

    Tommy-yw/RunbookHermes

    Pharmaceutical research assistant for drug discovery workflows.

    546 GitHub starsUsed in 1 repo~2.3k tokens
    Research & ScienceAuto-check passed
  • DiffDock Molecular Docking

    K-Dense-AI/scientific-agent-skills

    Predicts how small molecules bind to a protein with DiffDock, covering batch docking, pose ranking by confidence and checks on the results; not for binding affinity.

    48k GitHub starsUsed in 1 repo~3k tokens
    Research & ScienceAuto-check: notes
  • Routes bioinformatics, drug discovery, clinical and multi-omics tasks from a chat interface to Claude Code sessions running K-Dense scientific skills, with a live dashboard per task.

    617 GitHub starsUsed in 1 repo~2k tokens
    Research & ScienceAuto-check passed
  • Edu Chem Reaction

    wy51ai/edulab

    把一个化学反应做成自包含的微观 3D 交互演示网页:左/上为 Three.js 可交互分子动画 (拖滑块看断键·成键·原子重组,分步高亮),右为 KaTeX 反应方程 + 分步讲解 + 原子守恒计数 + 可选能量-反应进程曲线。支持三入口——给定文字反应/方程、随机出题、上传图片识别后演示。

    1.4k GitHub stars~1.2k tokensUpdated 9 days ago
    Research & ScienceAuto-check passed
  • Biopipelines

    locbp-uzh/biopipelines

    Design and run computational protein and ligand workflows on a GPU: binder and enzyme design, de novo backbone generation, inverse folding and sequence redesign, structure prediction, protein-ligand…

    109 GitHub stars~2.4k tokensUpdated 7 days ago
    Research & ScienceAuto-check passed

More from InternScience/scp

All 73 skills in this repo
  • Given an rsID, query multiple databases (dbSNP, FAVOR, GWAS Catalog, ClinVar, gnomAD, PharmGKB, ClinGen) for comprehensive annotation.

    169 GitHub starsUsed in 1 repo~2.2k tokens
    Auto-check passed
  • Drugsda Esmfold

    InternScience/scp

    Use ESMFold model to predict 3D structure of the input protein sequence.

    169 GitHub starsUsed in 2 repos~721 tokens
    Auto-check passed
  • Drugsda Prosst

    InternScience/scp

    Given a protein sequence and its structure, employ ProSST model to predict mutation effects and obtain the top-k mutated sequences.

    169 GitHub starsUsed in 2 repos~949 tokens
    Auto-check passed
  • Calculate atmospheric parameters including Coriolis parameter, geostrophic wind, heat index, potential temperature, and dewpoint for meteorology and climate science.

    169 GitHub starsUsed in 1 repo~2.1k tokens
    Auto-check passed
  • Biomedical Web Search

    InternScience/scp

    Search biomedical literature and web content using Tavily search engine for research and clinical information.

    169 GitHub starsUsed in 1 repo~598 tokens
    Auto-check passed
  • Calculate buoyancy forces and acceleration for fluid mechanics and hydrodynamics analysis.

    169 GitHub starsUsed in 1 repo~540 tokens
    Auto-check passed

Questions about Boltz2 Binding Affinity

What does Boltz2 Binding Affinity do?

Predict protein-ligand binding affinity using Boltz-2 model to assess molecular interactions and binding probability for drug discovery. Boltz2 Binding Affinity is an agent skill from InternScience/scp. Predict protein-ligand binding affinity using Boltz-2 model to assess molecular interactions and binding probability for drug discovery.

When should I use Boltz2 Binding Affinity?

Boltz2 Binding Affinity fits situations like: tasks that involve Drug discovery and cheminformatics.

How do I install Boltz2 Binding Affinity in Claude Code?

Run `npx skills add InternScience/scp --skill boltz2-binding-affinity -a claude-code`. Or copy the skill folder (skills/boltz2-binding-affinity in InternScience/scp) into .claude/skills/boltz2-binding-affinity in your project. Claude Code loads it when a task matches its description.

How do I install Boltz2 Binding Affinity in Codex?

Run `npx skills add InternScience/scp --skill boltz2-binding-affinity -a codex`. Or copy the skill folder (skills/boltz2-binding-affinity in InternScience/scp) into .agents/skills/boltz2-binding-affinity in your project. Codex loads it when a task matches its description.

Can I use Boltz2 Binding Affinity in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add InternScience/scp --skill boltz2-binding-affinity -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/boltz2-binding-affinity, .gemini/skills/boltz2-binding-affinity, .github/skills/boltz2-binding-affinity and .opencode/skills/boltz2-binding-affinity in your project.

What does Boltz2 Binding Affinity need to run?

SKILL.md names no scripts, command-line tools or credentials: Boltz2 Binding Affinity is instructions for the agent only. Our summary lists: Python 3.

Does Boltz2 Binding Affinity access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Boltz2 Binding Affinity safe to install?

Our automated static check of SKILL.md flagged 1 warning(s): contains a long base64 blob. Read the flagged lines before installing; the check is not a guarantee either way.

What licence does Boltz2 Binding Affinity use?

Boltz2 Binding Affinity is published under the MIT licence (declared in SKILL.md). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Boltz2 Binding Affinity use?

About 1.4k tokens (SKILL.md is roughly 5.8k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Boltz2 Binding Affinity?

Skills that share tags, products or a category with Boltz2 Binding Affinity: Molecode (AtomFlow-AI/MoleCode, 305 stars), Drug Discovery (Tommy-yw/RunbookHermes, 546 stars), DiffDock Molecular Docking (K-Dense-AI/scientific-agent-skills, 48k stars) and Biomedical Analysis Dispatch (xjtulyc/MedgeClaw, 617 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Boltz2 Binding Affinity?

InternScience (a GitHub organization) maintains it in InternScience/scp, which has 169 GitHub stars. The repository holds 73 skills in this directory. The repository was last updated on June 3, 2026.

Source: InternScience/scp on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.