Skills medchem
⚗️

medchem

Safe ⚙️ External commands📁 Filesystem access🌐 Network access

Filter molecules by drug-likeness rules

Also available from: davila7

Drug discovery requires filtering large compound libraries for drug-like properties. This skill applies established medicinal chemistry rules including Rule of Five, Veber, and PAINS filters to prioritize compounds and identify structural alerts efficiently.

Supports: Claude Codex Code(CC)
🥉 75 Bronze
1

Download the skill ZIP

2

Upload in Claude

Go to Settings → Capabilities → Skills → Upload skill

3

Toggle on and start using

Test it

Using "medchem". Filter this SMILES list for drug-likeness: CC(=O)OC1=CC=CC=C1C(=O)O, c1ccccc1, CCN

Expected outcome:

  • Aspirin: Passes Rule of Five
  • Benzene: Passes Ro5 but fails H-bond donor/acceptor requirements
  • Ethylamine: Passes Rule of Five
  • Summary: 2 of 3 molecules pass drug-likeness criteria

Using "medchem". Check for PAINS patterns in these molecules

Expected outcome:

  • No PAINS patterns detected in the first molecule
  • Second molecule contains rhodanine substructure (PAINS alert)
  • Third molecule flagged for catechol pattern

Security Audit

Safe
v4 • 1/17/2026

All 288 static findings are false positives. The static analyzer misidentified markdown code fences (backticks) as shell execution, Python file operations as Node.js fs, medicinal chemistry terminology (MD5, DES) as cryptographic weaknesses, and molecule validation as reconnaissance. This is a legitimate drug discovery library with no malicious code.

5
Files scanned
3,480
Lines analyzed
3
findings
4
Total audits

Risk Factors

⚙️ External commands (3)
SKILL.md:27-396 references/api_guide.md:12-581 references/rules_catalog.md:31-577
📁 Filesystem access (1)
scripts/filter_molecules.py:209-392
🌐 Network access (1)
SKILL.md:402-403
Audited by: claude View Audit History →

Quality Score

64
Architecture
100
Maintainability
85
Content
21
Community
100
Security
83
Spec Compliance

What You Can Build

Prioritize lead compounds

Filter virtual screening hits to identify drug-like candidates for synthesis.

Assess compound quality

Evaluate synthesized compounds against established medicinal chemistry rules.

Library design

Design focused compound libraries with optimal physicochemical properties.

Try These Prompts

Basic filtering 
Filter this list of SMILES using the Rule of Five and identify compounds that pass.
Structural alerts 
Check these molecules for PAINS patterns and common structural alerts. Report any alerts found.
Lead optimization 
Apply lead-like criteria and Lilly demerits filter to prioritize compounds for optimization.
Comprehensive analysis 
Apply multiple filters including Rule of Five, NIBR filters, and complexity metrics to my compound library.

Best Practices

  • Combine multiple filter types for comprehensive assessment
  • Use parallel processing for large compound libraries
  • Document filtering decisions for reproducibility

Avoid

  • Blindly accepting or rejecting compounds based solely on rule pass/fail
  • Ignoring structural alerts in favor of property-based rules alone
  • Applying strict filters too early in the discovery pipeline

Frequently Asked Questions

What is the Rule of Five?
Lipinski's Rule of Five predicts oral bioavailability based on MW ≤ 500, LogP ≤ 5, HBD ≤ 5, HBA ≤ 10.
What are PAINS filters?
Pan Assay Interference compounds are molecules that show activity through non-specific mechanisms.
Can I use this for covalent inhibitor design?
Yes, the skill can identify Michael acceptors and other reactive groups useful for covalent inhibitors.
What file formats are supported?
Input formats include CSV, TSV, SDF, and plain text files with SMILES.
Does this require RDKit?
Yes, medchem depends on RDKit for molecular operations. Install via pip install medchem.
How are results structured?
Results include pass/fail status for each rule, detailed violation information, and demerit scores.

Developer Details

Author

K-Dense-AI

License

Apache-2.0 license

Repository

https://github.com/K-Dense-AI/claude-scientific-skills/tree/main/scientific-skills/medchem

Ref

main

File structure

📁 references/

📄 api_guide.md

📄 rules_catalog.md

📁 scripts/

📄 filter_molecules.py

📄 SKILL.md