Pharmaclaw Tox Agent

Toxicology Agent for pharma drug safety profiling from SMILES. Computes RDKit ADMET descriptors (logP, TPSA, MW, HBD, HBA, rotatable bonds), Lipinski Rule of...

Security Scan

The skill's code, instructions, and declared dependency align with its stated purpose (RDKit-based SMILES toxicology profiling); it contains no apparent data-exfiltration, secret access, or unrelated capabilities.

Purpose & Capability

Name/description match the implementation: the SKILL.md and scripts/tox_agent.py perform RDKit descriptor calculations, Lipinski/Veber checks, QED, and simplified PAINS matching. Declared dependency (rdkit-pypi) is appropriate for the stated functionality and no unrelated credentials or binaries are requested.

Instruction Scope

The runtime instructions are narrowly scoped: run the provided Python script with a SMILES string. The SKILL.md does not instruct the agent to read unrelated files, access environment secrets, or call external network endpoints. The script only processes the input SMILES and returns a local JSON report.

Install Mechanism

No install spec is provided (instruction-only skill) which is low-risk. However, the declared dependency rdkit-pypi can be non-trivial to install (native/compiled components). The skill does not provide an automated install step—users must ensure RDKit is available in the runtime environment.

Credentials

No environment variables, credentials, or config paths are requested. The operations performed are local molecule analyses and do not require external secrets or system-level access.

Persistence & Privilege

always is false and the skill does not request persistent/system-level privileges. It does not modify other skills or system configuration.

Assessment

This skill appears internally consistent and implements what it claims: a small RDKit-based SMILES toxicology profiler. Before installing/running: (1) ensure RDKit (rdkit-pypi or a proper RDKit build) is available in an isolated environment — RDKit often requires compiled binaries; (2) be aware PAINS checking is a simplified/mock subset (SKILL.md notes this limitation), and the reported risk label is coarse ('Low' vs 'Medium/High'); (3) no network calls or secret access are present in the code, so there is no obvious exfiltration risk, but verify you run it in a controlled environment and review dependencies you install. If you need production-grade PAINS/QSAR/hERG or mutagenicity predictions, plan to replace the simplified implementations with validated models and a complete PAINS catalog.

PharmaClaw Toxicology Agent

Overview

Predictive toxicology and drug safety profiling agent for the PharmaClaw pipeline. Screens drug candidates via RDKit descriptors, rule-based filters (Lipinski Ro5, Veber), QED scoring, and PAINS alerts to flag safety risks early in the discovery process.

Quick Start

# Analyze a compound
python scripts/tox_agent.py "CC(=O)Nc1ccc(O)cc1"

# Default (ethanol)
python scripts/tox_agent.py

Capabilities

Check	Method	Threshold
Lipinski Ro5	MW, LogP, HBD, HBA	MW>500, LogP>5, HBD>5, HBA>10
Veber Rules	TPSA, Rotatable Bonds	TPSA>140, RotB>10
QED Score	RDKit QED module	0-1 (higher = more drug-like)
PAINS Alerts	Substructure matching	Known assay interference patterns
Ring Analysis	Aromatic/total ring count	Complexity indicator

Decision Tree

• SMILES input → RDKit descriptor calculation → Rule-based screening
• Lipinski violations = 0 AND PAINS = 0 → Risk: Low
• Any violations or PAINS hits → Risk: Medium/High
• High risk flagged → Recommend analogs via Chemistry Query / IP Expansion

Output Format

{
  "lipinski_viol": 0,
  "veber_viol": 0,
  "qed": 0.737,
  "pains": 0,
  "risk": "Low",
  "props": {
    "mw": 151.2,
    "logp": 1.02,
    "tpsa": 49.3,
    "hbd": 2,
    "hba": 2,
    "rotb": 1,
    "rings": 1,
    "arom": 1
  }
}

Risk Classification

• Low: No Lipinski violations, no PAINS alerts, QED > 0.5
• Medium: 1-2 Lipinski violations OR low QED
• High: 3+ Lipinski violations, PAINS hits, or multiple Veber violations

Chain Integration

• Receives from: Chemistry Query (SMILES), Pharmacology (ADME flags)
• Feeds into: IP Expansion (safer derivative suggestions), Synthesis (avoid toxic intermediates)
• Cross-references: Market Intel (FAERS adverse events for similar structures)

Dependencies

• rdkit-pypi — Molecular descriptors, QED, substructure matching

Scripts

• scripts/tox_agent.py — Main agent: ToxAgent class with analyze(smiles) method

Limitations

• PAINS screening uses simplified substructure set (production should use full PAINS catalog)
• No Ames mutagenicity or hERG channel prediction (descriptor-based proxies planned)
• LD50 estimation not yet implemented (QSAR model planned for future version)