Tutorial
10-part hands-on guide — from first annotation to full generation pipeline.
@libar-dev/delivery-process
Code-first documentation and delivery process toolkit. Extract patterns from TypeScript and Gherkin sources, generate living documentation and architecture diagrams, and enforce delivery workflow via FSM.
Why This Exists
Section titled “Why This Exists”Traditional documentation fails because it exists outside the code. When AI agents receive unstructured markdown, they hallucinate APIs, ignore architectural constraints, and consume excessive token budgets.
This package inverts the paradigm: the Git repository is the event store, and all generated documentation is a derived projection of annotated source code.
Quick Start
Section titled “Quick Start”npm install @libar-dev/delivery-processpnpm add @libar-dev/delivery-processExplore
Section titled “Explore”Guides
Methodology, configuration, session workflows, Gherkin patterns, annotation guide.
Reference
Architecture deep-dive, Data API CLI, Process Guard, validation rules, taxonomy.
Product Areas
Auto-generated docs with Mermaid diagrams, extracted types, and behavior specs.
Key Capabilities
Section titled “Key Capabilities”| Capability | Description |
|---|---|
| Pattern Extraction | TypeScript JSDoc and Gherkin tag scanning with shape extraction |
| MasterDataset | Single read model with O(1) pre-computed views |
| Codec Rendering | 20 codecs generating markdown with progressive disclosure |
| FSM Workflow | Finite state machine enforcement via pre-commit hooks |
| Data API CLI | Query delivery state directly — 5-10x less context than raw file reads |
| Dual-Source Architecture | TypeScript owns implementation tags, Gherkin owns planning tags |