MolVision Molecular Image Recognition (OCSR) for Openclaw

A specialized skill for converting molecular structure images into digital chemical data and calculated properties using PatSight MolVision.

snailwhb
v1.0.1
Mar 19, 2026
0
794
0

Install & Download

1. ClawHub CLI

The fastest way to install a skill directly from the registry.

npx clawhub@latest install ocsr

2. Manual Installation

Copy the skill folder to one of these locations

Global
~/.openclaw/skills/
Workspace
<project>/skills/

Priority: Workspace > Local > Bundled

3. Prompt Installation

Copy this prompt to OpenClaw to install it automatically.

Help me install ocsr using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).

Prefer to download?

Get the raw skill files in a ZIP archive.

What is MolVision Molecular Image Recognition (OCSR)?

The MolVision Molecular Image Recognition skill is a professional-grade tool designed for Optical Chemical Structure Recognition (OCSR). It leverages the PatSight Patent Extractor API to transform raw images of chemical structures into standardized digital formats like SMILES and SDF. By integrating this functionality into Openclaw Skills, researchers and developers can automate the transition from visual diagrams to actionable chemical informatics data.

Beyond simple recognition, this skill enriches the extracted data using RDKit to calculate critical molecular properties such as Molecular Weight, LogP, and TPSA. It provides a complete workflow for digitizing chemical information found in patents, papers, or lab notebooks, resulting in both machine-readable JSON data and human-readable visualization dashboards.

MolVision Molecular Image Recognition (OCSR) Use Cases

  • Digitizing chemical structures found in patent filings or scientific publications.
  • Batch processing legacy image folders to build searchable chemical databases.
  • Automating the calculation of molecular properties directly from visual structure inputs.
  • Generating visualization reports for chemical datasets derived from image recognition.

How MolVision Molecular Image Recognition (OCSR) Works

  1. The user inputs one or more molecular images in formats such as PNG, JPG, or TIFF.
  2. The skill authenticates with the PatSight service using an account, password, or session token.
  3. The images are processed via the PatSight MolVision engine to perform optical recognition of chemical bonds and atoms.
  4. The engine returns SMILES strings and SDF data, which are then passed to RDKit for property enrichment.
  5. The skill generates a structured JSON output containing all chemical metadata and a dashboard image for visual verification.

MolVision Molecular Image Recognition (OCSR) Setup

To use this skill within Openclaw Skills, you must first register for an account at PatSight.

Configure your environment variables:

export PATSIGHT_ACCOUNT="your_account"
export PATSIGHT_PASSWORD="your_password"

Run the recognition script using uv:

uv run python skills/ocsr/scripts/run_ocsr.py your_molecule.png --outdir ./output

MolVision Molecular Image Recognition (OCSR) Data Schema & Taxonomy

The skill produces two primary output files for every execution:

File Type Description
ocsr_result_<timestamp>.json Contains raw API responses, SMILES, SDF strings, and RDKit-calculated properties (MW, LogP, TPSA, Formula).
ocsr_dashboard_<timestamp>.png A visualization report containing the recognized structure diagrams.

Each entry in the JSON output includes a recognition score to indicate the confidence of the OCSR process.

MolVision Molecular Image Recognition (OCSR) Advanced Features

  • Multi-image batch processing allows for 1-N files to be analyzed in a single command execution.
  • Integrated RDKit fallback ensures that even if property parsing fails, the raw structural data is preserved.
  • Flexible authentication priority supporting command-line tokens, environment variables, or account credentials.
  • Customizable dashboard reporting with configurable panel image sizes for high-resolution visualizations.

SKILL.md


Loading

Related Openclaw Skills

METADATA

Requires
Bins uvpython
Github Stars: 0
forks: 0

Featured*