Adme Property Predictor
AdvisoryAudited by Static analysis on Apr 30, 2026.
Overview
No suspicious patterns detected.
Findings (0)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
If invoked broadly, the agent could run local commands or modify files while performing ADME analysis.
The skill permits local file and shell-related tools. This is plausibly needed to run the predictor and write outputs, but it is broader than pure in-memory calculation.
allowed-tools: [Read, Write, Bash, Edit]
Use the skill in a project directory, review proposed Bash/Edit actions, and keep outputs limited to files you explicitly choose.
Results or runtime behavior could vary depending on which dependency versions are installed.
The dependencies are unpinned, so an installation may resolve to different package versions over time. RDKit is purpose-aligned for cheminformatics, and no malicious install behavior is shown.
dataclasses rdkit
Install dependencies from trusted sources and prefer pinned versions in a controlled environment for reproducible scientific work.
Users could overestimate the scientific reliability of the predictions if they treat them as validated pharmacokinetic results.
The visible implementation describes simplified heuristic models, while the documentation presents the skill as using validated cheminformatics models and ML/QSPR-style predictions. The documentation also includes warnings against clinical dosing use, which reduces but does not remove the need for caution.
# Simplified model: high lipophilicity, moderate PSA favors permeability
Treat outputs as preliminary screening estimates only, and validate important ADME conclusions with experimentally supported or independently validated models.