AutoBiology CLI

AutoBiology turns biological SOP text into structured automation requirements. The CLI parses procedure text, builds operation hypergraphs, generates deterministic R1-R10 requirements, optionally adds LLM-assisted candidates, and writes coverage reports with Mermaid diagrams.

Quick Start

npm install -g autobiology-cli
autob --help
autob update --check
autob init
autob run your-sop.txt -o out

Open out/report.md first, then inspect the JSON files for structured data. The installed command is autob.

Knowledge-Aware Pipeline

The published CLI ships with the curated knowledge base in data/:

• synonyms.json normalizes biological materials, containers, tools, locations, and common units.
• domain-patterns.json defines required parameters, risks, requirement links, and inference hints by action.
• parameter-constraints.json records common units, ranges, thresholds, and tolerances.
• risk-catalog.json provides standard risk handling and verification methods.

autob run loads this knowledge base once and passes it through atomization, hypergraph construction, and deterministic requirement generation. LLM calls remain optional and are only used for candidate requirements when configured.

Research Artifacts

• publication/ contains manuscript-oriented method notes, reproducibility instructions, sample outputs, Mermaid diagrams, and a draw.io-compatible algorithm figure.
• graphify-out/ contains the repository knowledge graph, architecture report, and navigable graph HTML for codebase review.

Documentation

• Hosted docs: https://cbx-2023.github.io/AutoBiology/
• 中文文档: https://cbx-2023.github.io/AutoBiology/zh/
• Getting Started
• Blender Simulation
• Configuration
• Pipeline Guide
• LLM And Artifacts
• CLI Reference

To confirm whether LLM inference is enabled, run autob config show. After a run, inspect run-meta.json, 06-clarifications.json, and any LLM-Candidate entries in 04-requirements.json.

autob config show
node -e "const fs=require('fs'); const m=JSON.parse(fs.readFileSync('out/run-meta.json','utf8')); console.log(m.config.llmModel)"
rg -n '"LLM-Candidate"' out/04-requirements.json

not-configured and no LLM-Candidate matches means the run used the deterministic algorithm layer only.

Blender Simulation

The Blender simulation code is a standalone component in components/blender-simulation. It consumes generated requirement JSON and can emit a Blender Python script that renders MP4 output through Blender's native FFmpeg runtime. It is intentionally not wired into the original SOP converter pipeline.

Commands

autob --help
autob init
autob config show
autob update --check
autob update
autob run <sop-file> -o <output-dir>
autob atomize <sop-file> -o <output-dir>
autob hypergraph <op-table> -o <output-dir>
autob requirements <nodes-file> <hyperedges-file> -o <output-dir>
autob infer <requirements-file> -o <output-dir>
autob review <requirements-file> -o <output-dir>

Development

npm test
npm run build

Node.js 20+ is required.