Getting Started
This guide gets AutoBiology running locally in about five minutes.
Install
npm install -g autobiology-cli
autob update --check
For source checkout development:
npm install
npm run build
npm link
For local development you can also run the TypeScript source with npx tsx src/cli.ts. After installing or linking the package, use the autob executable.
Use autob update to install the latest published CLI later. Automatic update checks are cached for 24 hours; set AUTOB_DISABLE_UPDATE_CHECK=1 when running in CI or offline shells.
Initialize Configuration
Run the setup wizard:
autob init
The wizard writes ~/.autob/config.json. It asks for the LLM provider, API key, model, timeout, and optional connectivity check. The pipeline still works without a configured LLM; it will use deterministic requirement generation and record an LLM fallback clarification.
Run The Pipeline
autob run tests/fixtures/sample-sop-cell-collection.txt -o out
The output directory contains:
01-ops.json
02-nodes.json
03-hyperedges.json
04-requirements.json
05-coverage.json
06-clarifications.json
diagrams/
report.md
run-meta.json
Start with out/report.md. Use out/04-requirements.json for structured requirements and out/05-coverage.json for requirement coverage by hyperedge.
What The Built-In Knowledge Base Does
The npm package includes the curated data/ files used by the deterministic pipeline. They normalize SOP aliases such as PBS buffer, Falcon tubes, biosafety cabinets, and common unit spellings; add action-specific default tools, risks, and output states; and create missing-parameter clarifications from domain patterns.
No LLM configuration is required for this behavior. LLM providers are only used for optional candidate requirement inference.
To confirm whether LLM inference is active, run autob config show and inspect out/run-meta.json plus out/06-clarifications.json after a pipeline run. See LLM And Artifacts for details.
Repository Reference Artifacts
When working from the GitHub repository, publication/ provides paper-style method notes, reproducibility instructions, sample run outputs, Mermaid diagrams, and a draw.io-compatible algorithm diagram. graphify-out/ provides the generated code knowledge graph and architecture report for codebase review.
Useful Next Commands
autob config show
autob update --check
autob run your-sop.txt -o out --interactive
autob review out/04-requirements.json -o out --interactive