# Jerry Orta

> The platform layer that keeps front-ends coherent as they scale: design systems, Nx monorepos, and data-visualization tooling — with the source behind them.

Each Blog and Concepts link is the page's Markdown source, with front matter removed and links made absolute; the HTML page is the same URL without `.md.txt`.

## Blog

- [What the Repository Remembers](https://jerryorta.com/blog/what-the-repository-remembers.md.txt): Three registers of context an AI agent reads — normative documentation tracked with the code, observational memory in a repository of its own, and the plans that state intent — and why keeping them apart is what makes complicated work both cheap and accurate.
- [Charts as a Library, Not a Component](https://jerryorta.com/blog/charts-as-a-library.md.txt): Why the charting layer is a config-driven, layered library I own — not a third-party component I rent — and why an agent-extensible shape makes owning it cheap to scale.

## Concepts

- [Domain-Driven Design](https://jerryorta.com/concepts/domain-driven-design.md.txt): How a large front-end organizes itself around bounded contexts, a ubiquitous language, and self-contained domains — each domain owning its models, state, UI, and tokens as one unit, grounded in a real personal-finance domain built on a shared foundation.
- [The Claude Code Skills I Use](https://jerryorta.com/concepts/claude-skills.md.txt): The custom Claude Code skills that encode my workspace conventions — scaffolding, state placement, theming, and verification.
- [Charts Library Architecture](https://jerryorta.com/concepts/charts-library-architecture.md.txt): The mechanics behind the config-driven chart library — layer composition on a shared base, per-import tree-shaking, and an Angular tooltip positioned by D3 outside change detection.
- [Design Library Architecture](https://jerryorta.com/concepts/design-library-architecture.md.txt): How a themeable component library holds together — a design library and a themes library joined only by a CSS-variable contract, so one component set wears many skins, restyles without recompiling, and a domain can layer its own tokens on top.
- [Table Library Architecture](https://jerryorta.com/concepts/table-library-architecture.md.txt): The mechanics behind a data table built on a headless engine — a locked three-lane substrate, four extension seams proven by construction rather than by review, state the host owns so a view can be saved and restored, and marks stored as descriptors instead of enumerations.
- [The Agent-Files Architecture](https://jerryorta.com/concepts/claude-agent-files.md.txt): How a large monorepo makes itself legible to an AI agent — a single always-loaded core, colocated AGENTS.md pointers, a routed deep-dive library, and longer-lived project memory, each loading only when it is relevant.
- [Framework Source as Agent Reference](https://jerryorta.com/concepts/oss-clones-ai-reference.md.txt): How an AI agent stays current with a fast-moving framework — a read-only, version-pinned set of upstream source clones kept beside the repository and synced by one command, so latest-pattern work reads what the framework does now rather than what a model remembers.
- [From Backlog to Merged PR](https://jerryorta.com/concepts/epic-workflow.md.txt): How a feature moves from a backlog item to a merged pull request when the tracker and the content are kept apart — status in Jira, the plan in a version-controlled file per ticket, and a family of /epic-* skills that read both, do the work, and write the status back.

## Optional

- [Design library](https://jerryorta.com/libraries/design-library): One primitive set, composed into two product domains
- [Charts](https://jerryorta.com/libraries/charts): 33 chart types on one configuration contract
- [Table](https://jerryorta.com/libraries/table): Spreadsheet-grade editing on a virtualized grid
- [Calendar](https://jerryorta.com/libraries/calendar): Five views, drag-and-resize, and typed event payloads