A self-taught developer from Bangalore. ToolForge is one of the things I've built. Here is the story behind it.
I'm a self-taught developer based in Bangalore, and I spend most of my free time building small web tools that scratch my own itches. A free resume builder, an AI career-path helper, an SEO toolkit, and a handful of others. ToolForge is one of those projects.
It started because I wanted one place to keep track of AI tools and the tech I use, instead of digging through bookmarks, newsletters, and half-remembered Reddit threads. So I built a list. The first version was a single hand-coded HTML file with no build step, just data crammed into a 7.11 MB document.
It did the job, but it was slow to open, a pain to update, and impossible to share. So I rebuilt it properly, first with Vite and now on Next.js and React, splitting the data into chunks that load on demand and building the search index ahead of time.
It now holds 2,386 AI tools across writing, coding, image generation, video, SEO, and a few dozen other categories, plus 1,861 developer technologies covering most of the stack. I add and check entries by hand. Nothing scraped, nothing paid for a spot.
The idea hasn't changed: a quick reference I'd actually trust. Search is fuzzy because I make typos. There's a comparison view because picking between five similar tools is genuinely annoying. You can export your favorites because your shortlist shouldn't be locked inside someone else's app.
It's free and it'll stay that way. No account, no limits, no upsell. If it's useful, pass it to someone. You can find more of my work on GitHub, see my portfolio, or say hi on Twitter.
A modern React stack on the front, a fast hand-rolled catalog engine underneath.
App Router, React Server Components, and an optimized production build. One dev server, proper routing, no more file:// gotchas.
The marketing pages are React components; the hub mounts its catalog logic in a client component with hooks for lifecycle.
A custom dark design system with CSS variables, no UI kit. Gradients, glass, and animation built from scratch.
Custom Levenshtein edit-distance matching with synonym expansion and a pre-compiled index. No external search library.
Favorites, density, and last-visited section persist across sessions, with graceful fallback when storage is blocked.
9 MB of catalog data loads on demand from /public, so the initial page stays light and fast.
The rules I keep coming back to when deciding what goes in.
Data loads on demand, assets are split, and the search index is built ahead of time so nothing is crunched while you wait.
No login, no limits, no paywall. You shouldn't have to hand over an email just to look something up.
I'd rather have 2,386 tools I've actually checked than 20,000 scraped entries full of dead links and duplicates.
Keyboard navigation throughout, with proper ARIA roles, focus handling, and skip links, not bolted on as an afterthought.
4,247 tools and technologies, one fast catalog. No account needed.