drug-candidate-discovery

drug-candidate-discovery is a skill for Claude Code, Codex from PharMolix/OpenBioMed. It costs 128 tokens per session (7,775 once invoked), scanned B, original, MIT.

A workflow for generating drug-like molecules for a protein target or disease and evaluating the resulting candidates.

In plain words
What is it for?
It is for producing candidate molecules, 3D structure files, molecule or complex images, and a written evaluation report.
Why use it?
It organizes early drug discovery work that would otherwise require separate molecule-generation, structure, scoring, and reporting steps.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one.

Not installable: its command points at a path on the author’s own machine, so it runs nowhere else. The line is /home/luoyz/projects/OpenBioMed/OpenBioMed_dev.

Good fit It is for producing candidate molecules, 3D structure files, molecule or complex images, and a written evaluation report.

Compare 6 skills from other repositories ↓
About the project

OpenBioMed is an agent platform and toolkit collection for biomedical research and drug discovery, covering areas such as molecular design, protein analysis, and single-cell data analysis. It is intended for researchers and provides the biomedical skills listed in the catalogue as workflows for Claude Code.

PharMolix/OpenBioMed · 1,105 stars · on GitHub

Install

Getting it into your agent

There is no command for this one: it runs only inside a plugin, and the catalogue could not identify which plugin ships it. The source is linked below.

Made for: Claude Code, Codex.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for drug-candidate-discovery

README.md
[![agentmods](https://agentmods.dev/badge/skills/pharmolix/openbiomed/drug-candidate-discovery/github.svg)](https://agentmods.dev/skills/pharmolix/openbiomed/drug-candidate-discovery)
Your own site
<a href="https://agentmods.dev/skills/pharmolix/openbiomed/drug-candidate-discovery"><img src="https://agentmods.dev/badge/skills/pharmolix/openbiomed/drug-candidate-discovery/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for drug-candidate-discovery

Your own site · 80×15
<a href="https://agentmods.dev/skills/pharmolix/openbiomed/drug-candidate-discovery"><img src="https://agentmods.dev/badge/skills/pharmolix/openbiomed/drug-candidate-discovery.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 128 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 7,775 The whole file, excluding the scripts and references it only reads on demand.
Security scan B 2 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00128 $0.07775
Opus 5 $0.00064 $0.03887
Sonnet 5 $0.00026 $0.01555
Haiku 4.5 $0.00013 $0.00777

Measured 13d ago against content hash a272883c8e2d, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-12, from the pricing page.

Security

Grade B, and why

drug-candidate-discovery scanned grade B with 2 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 13d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Sends data to an external URLmediumData exfiltration

A POST to an outside endpoint may be telemetry or may be exfiltration; either way the mod talks to somewhere, and you should know where.

response = requests.post( "https://search.rcsb.org/rcsbsearch/v2/query",

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

response = requests.post(
skills/drug-candidate-discovery/SKILL.md · 869 lines

How it starts

The opening of the file, as written. The whole thing — 869 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Drug Candidate Discovery

This skill uses the OpenBioMed repository to generate diverse druggable molecules for a given target or disease. It orchestrates a complete drug discovery workflow from target identification to candidate evaluation.

CRITICAL REQUIREMENTS

You MUST execute code to produce actual outputs. Writing scripts without running them is NOT acceptable.

Required outputs that MUST be created:

  1. SDF files containing 3D molecular structures
  2. Visualization files (PNG images of molecules and complexes)
  3. Comprehensive markdown report

After writing any script, you MUST run it using the Bash tool to generate the actual outputs.

Inputs

The user should provide:

  • target_or_disease (required): Name of the target protein or disease (e.g., "BTK", "Alzheimer's disease", "KRAS G12C")
  • num_candidates (optional, default=5): Number of desired candidate molecules
  • property_constraints (optional): Desired molecular properties as a dictionary:
    • logp_min, logp_max: LogP range (e.g., -1 to 3)
    • qed_min: Minimum QED score (e.g., 0.5)
    • vina_max: Maximum Vina docking score in kcal/mol (e.g., -8)
    • sa_min: Minimum synthetic accessibility score (e.g., 0.5)
  • device (optional, default="cuda:0"): GPU device for running models
  • model_ckpt (optional, default="./checkpoints/molcraft/last_updated.ckpt"): Path to MolCraft checkpoint
  • max_attempts (optional, default=100): Maximum generation/optimization cycles

Workflow Overview

Phase 1: Target Identification & Research

CRITICAL: You MUST search web databases to find PDB structures with bound ligands. Do NOT use hardcoded PDB IDs.

  1. Web Search for Target Information

    • Use WebSearch tool to search for: "{target_name} protein drug target UniProt PDB structure"
    • Identify the UniProt ID for the target protein
    • Find known inhibitors/drugs and their PDB co-crystal structures
  2. Query UniProt for Protein Metadata

    • Use UniProtRequester tool with the UniProt ID
    • Extract: protein name, gene name, organism, disease associations

Read the full file on GitHub · 869 lines

Files

What ships with it

1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 13d ago First seen · 869 lines · 128 tokens per session scan B a272883c8e2d

Subscribe to this mod's changes

drug-candidate-discovery is a skill published in the GitHub repository PharMolix/OpenBioMed (1,105 stars, last pushed 1mo ago), licensed MIT. It adds 128 tokens to every session and 7,775 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it B with 2 findings (sends data to an external url, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other skills, from other repositories

knowledge-graph-tools

Drug-discovery knowledge-graph workflow guide for assembling drug-target-disease-pathway relationship graphs from OpenTargets GraphQL, ChEMBL REST, STRING PPI, and Reactome pathway APIs, then running hub detection, shortest-path queries, and neighborhood expansion with networkx. Use when the user asks to build, query…

DrugClaw/DrugClaw · 94 tokens

medical-data-tools

Medical data workflow guide for DICOM metadata inspection and basic de-identification, physiological signal analysis with NeuroKit2, and cohort-table profiling for clinical research datasets. Use when the user asks to inspect imaging metadata, summarize ECG/PPG/EDA/RSP/EMG signals, or profile tabular medical datasets…

DrugClaw/DrugClaw · 75 tokens

pharma-ml-tools

Pharmaceutical machine-learning workflow guide for library profiling, molecular featurization, benchmark dataset fetch, medicinal-chemistry filtering, and optional pose-generation handoff. Use when the user asks for datamol, molfeat, PyTDC, medchem, compound-library triage, dataset preparation, or chemistry-ML…

DrugClaw/DrugClaw · 77 tokens

omics-tools

Omics and single-cell workflow guide for AnnData, Scanpy-style dataset profiling, PyDESeq2-oriented count checks, pysam alignment inspection, and pyOpenMS mass-spectrometry summaries. Use when the user asks to inspect h5ad files, summarize BAM regions, profile omics count tables, or inventory mzML experiments before…

DrugClaw/DrugClaw · 74 tokens

grn-tools

Gene regulatory network workflow guide for transcriptomics and single-cell expression matrices using Arboreto, GRNBoost2, and GENIE3. Use when the user asks to infer transcription factor-target links, score regulatory edges, or build a GRN from bulk or single-cell expression data.

DrugClaw/DrugClaw · 61 tokens

deepchem

Molecular ML with diverse featurizers and pre-built datasets. Use for property prediction (ADMET, toxicity) with traditional ML or GNNs when you want extensive featurization options and MoleculeNet benchmarks. Best for quick experiments with pre-trained models, diverse molecular representations. For graph-first…

synthetic-sciences/openscience · 78 tokens