Hi everyone, I'd like to share a developer tools site I've been polishing lately: KitVerse.dev.
Why I built this
AI seems capable of everything these days, but day-to-day AI work still comes with plenty of headaches:
- AI is so powerful — but what can we actually do with it? What products are other people building with it?
- Tokens are expensive. If you don't test your freshly issued API key yourself first, your AI agent goes full throttle — how much is that going to cost?
- Heard that thinking like a boss and "PUA-ing" your AI employee makes it work harder — is that actually true?
- Skill packages are everywhere these days. Has anyone audited their security? Are they going to stab me in the back?
Meanwhile, most online tool sites share a few common problems: they force you to sign up, upload whatever you paste to their servers, or hit you with a paywall after two free uses.
At the end of the day, real safety means keeping your data in your own hands. KitVerse's approach is local-first: every tool runs entirely in your browser — no signup, works offline, and nothing you paste is ever sent to a server or fed into analytics events.
AI tools first
Unlike traditional tool sites, KitVerse focuses on AI / LLM workflows. Here are a few highlights I'm particularly happy with:
Skill Security Scanner. Agent skill packages are flying around everywhere right now, and installing a SKILL.md from an unknown source is genuinely risky. This tool statically scans SKILL.md files, scripts, folders, ZIP packages, and even online skill links for malicious execution, data exfiltration, credential theft, and prompt injection risks. Scan before you install — peace of mind is worth it.
PUA AI Generator. Yes, it literally means PUA-ing your AI (lol). It generates "pressure prompts" and complete skill files that force AI coding agents to exhaust every option before giving up — the perfect cure for agents that try twice and then declare "I can't do it." Chinese is supported, with a choice of big-tech flavors.
API Playground. Got a new API key? Don't hand it to your AI to burn through just yet — come test it here yourself first. No need to install Postman or any other client: test REST and GraphQL APIs right in your browser, and debug WebSocket, SSE, Socket.IO, and MQTT connections directly, with requests and responses crystal clear. Get to know your API's temperament and cost before deciding whether to let an agent loose on it.
Daily New Words. The AI world coins new terms at a dizzying pace. This delivers a daily digest of new terms and concepts worth knowing — perfect for staying in sync over a coffee break.
Beyond these, the AI tool lineup includes: Tokenizer Playground (see exactly how text gets tokenized and compare costs across models), RAG Chunking Playground (preview 6 chunking strategies with real token counts and source offsets), a prompt toolchain (diff & compare, structured builder, batch variable expansion), a full Skill toolkit (validate, package, generate install commands, convert prompts to skills), plus practical data tools like model pricing & context comparisons and API provider compatibility comparisons.
Traditional tools on the side
All the usual developer tools are here too: JSON / YAML / TOML conversion, regex testing, encoders & decoders, hashing & crypto, UUID / password generators, image tools, timestamp conversion, cron parsing… 100+ of them. It works perfectly fine as a more private traditional tool site, too.
Some details
The site is pure frontend — inputs are never uploaded or saved. Token counts are marked "exact" for verified tiktoken mappings and clearly labeled "estimate" otherwise — no fudging. Model pricing data comes from LiteLLM's public snapshot, with source and date shown for every entry; the providers' official billing pages always have the final say.
Feedback welcome
The site is still being actively updated. All feedback is welcome: which tools are awkward to use, what's missing, any false positives in the scanner — I'd love to hear it all.
Top comments (0)