Case Study

AI-Dekrov

A public, moderated catalog of AI tools: services, coding agents, models, CLIs, desktop apps, and platforms - organized without the noise.

Launched Static GitHub Pages Moderated catalog Search + Filters

Open source

Browse the catalog or inspect the code

Every tool gets a dedicated page with official links, supported platforms, pricing, and install commands. The public data and source live in an open repository.

Open live site View on GitHub

Result

AI-Dekrov shows how a curated product can stay simple on the surface while having real structure underneath: a privacy-friendly static frontend, a publicly versioned data store, and a moderated contribution flow backed by automation. As a portfolio case it demonstrates product thinking around trust, maintainability, and community input - not just another list of links.

Problem

AI tools are scattered across newsletters, bookmarks, and ad-driven lists. Finding a trustworthy service, model, or CLI usually means filtering through noise, outdated links, and vague marketing instead of a compact, factual reference.

Solution

AI-Dekrov is a single, community-shared catalog where every entry has a verifiable official source, supported platforms, pricing, install commands, and a dedicated page - with unknown information left empty instead of invented.

Full-text search Search by name, description, tag, category, platform, or model.

Smart filters Filter by price and platform: Web, Desktop, CLI, VS Code, and API.

Your browser Favorites and theme preference are saved locally and never leave the browser.

Per-tool pages A dedicated page for every tool, with its official links and commands.

Features

Explore without noise

Trusted links, supported platforms, pricing, and short factual descriptions are presented together, so you can compare options fast and decide.

Personal, private

Save favorites and your theme directly in the browser. Personal notes stay local and are never included in public data or submissions.

One page per tool

Open a dedicated page for any entry to see its official links, platforms, pricing model, install commands, and the facts behind it.

How tools are added

Moderated submissions

Anyone can suggest a tool. Submissions are checked for valid JSON, URLs, enums, IDs, and duplicate entries, then a moderator reviews the result before anything is published. Nothing appears automatically.

Three entry paths

Smart Add starts from an official URL and auto-builds a schema-valid record in GitHub Actions. Manual is a full editor, and JSON Import lets you validate schema-ready records before submitting for review.

Smart Add, Manual, and JSON Import produce the same kind of checked, moderated submission.
For deeper local work, the owner can run a CLI that downloads a page, shows a preview, and writes to the data file only after confirmation.
The public catalog lives in the repository and is identical for every visitor.
Entries describe real tools with verifiable official sources; unknown information stays empty rather than invented.

Technology Stack

HTML + CSS + JS Single-page catalog UI: search, filtering, favorites, and per-tool pages, all client-side. Git + GitHub Actions Versioned public data and automated validation of tool submissions. JSON data Catalog entries, labels, and site config stored as plain JSON in the repository. Node.js tooling CLI scripts and an analyzer that help inspect pages and build tool records. GitHub Pages Static hosting with a custom domain - fast, cacheable, and dependency-free.