Software Engineering Specialist at TCS, based in Bhubaneswar, India. ML enthusiast, agent-tooling hacker, and the solo engineer behind the free public tools at oriz.in.

I ship products: web apps, public APIs, browser extensions, mobile apps, VS Code extensions, and AI / agent tooling. Pure-JS where it fits, Cloudflare at the edge where it deploys, MIT-licensed across the board. No signup walls, no ads inside the tools, no card-on-file required to use anything I publish.

• The oriz.in family. A growing set of free, public, privacy-respecting web tools and reference APIs at *.oriz.in. Each one solves a problem I personally hit and couldn't find a clean free alternative for.
• Agent tooling. Reusable Claude Code skills, MCP servers, and reusable CI workflows that make AI coding agents more useful end-to-end. Everything's Dagger-portable — same pipeline runs locally, on GH Actions, GitLab, and Codeberg.
• Browser + VS Code extensions. Small, focused tools that take the friction out of common workflows — bookmarks, secrets, clickbait, rewriting.

📫 chirag127.in · linkedin.com/in/chirag127 · @singhal127

A family of free public tools. Astro 6 + React 19 + Tailwind v4 on the frontend, Cloudflare Pages + Workers at the edge, Firebase Spark where state is needed. Common design language across every property, shared component library, and a single OKF knowledge bundle in the workspace umbrella that documents the architecture decisions.

The shape:

• Apps are static-rendered Astro with React islands where interactivity earns its keep.
• APIs are pure Cloudflare Workers — no Node servers, no cold starts, no card.
• Books are Markdown manuscripts built with an in-house Astro pipeline that compiles to web + EPUB + PDF.

Free, no-auth, CORS-open. Deployed on Cloudflare Workers or GH Pages.

Source manuscripts for the oriz book series — written in Markdown, built with a custom Astro-based pipeline.

• chirag127/100-year-strategy-book — 100-year personal-strategy book.
• chirag127/janaushdhi-book — companion book to the Janaushdhi medicine catalog.

• chirag127/sops-lens-vsc-ext — VS Code extension. Renders SOPS-encrypted values inline as CodeLens / hover / ghost-text. Decrypts in-memory; plaintext never touches disk.
• chirag127/userscripts — Tampermonkey / ScriptCat collection. Includes serp-open-articles and dearrow-show-original.
• chirag127/agent-skills — reusable skills for Claude Code and other agent harnesses. Junctioned into ~/.claude/skills/ and per-agent skill dirs.
• chirag127/workflows — Reusable GitHub Actions workflows for the whole fleet. Every workflow calls Dagger; downstream repos consume with 5-line pins.
• chirag127/ddgs-mcp-server — DuckDuckGo MCP server for AI agents. Drop-in search backend for any agent harness that speaks MCP.

A short list of the rules every repo follows. They keep the family coherent across a long tail of small projects.

• Free at the point of use. No signup walls in the app shell. No ads inside the tools. Donation links in the footer, never in the way.
• No card on file. Every backing service either has a permanent free tier I can stay inside, or I don't use it.
• MIT-licensed. Fork it, mirror it, run it yourself. The hosted version is a convenience, not a gate.
• Privacy by default. Where data lives client-side, it stays client-side. Where E2EE is feasible, it's optional and on by default.
• Brand-independent naming. Repo names describe the FUNCTION, not the brand — makes future domain migration cheap.

The personal-namespace work — experiments, full-stack web apps, mobile apps, and AI tooling. A mix of polished products and idea-validation prototypes.

• code-cringe-detector — lint for AI-written code smells (the tells that reveal a generation was machine-authored without a human pass).
• CogniPilot — LLM-driven desktop automation engine; mimics human computer interaction to chain real-world workflows.
• ClaudeBridge — high-performance Python CLI for resilient Claude interaction with structured output parsing and prompt orchestration.
• prompts — curated AI prompts for architects and DevOps. Battle-tested system prompts I reach for repeatedly.

• SynthSum — AI video summarizer (iOS / Android / Web). Text-to-speech with word highlighting, summary history, dark mode.
• SkillSwap — community barter network with time bank, verified reviews, Node / Express backend.
• ScanFlow — high-speed QR scanner + manager.

• ShopFlow — grocery e-commerce in Java 17 / Spring Boot / React / PostgreSQL. JWT auth, dual customer + admin UIs, Dockerized.
• OmniCampaign — enterprise ad-campaign management across major ad platforms from a single dashboard.
• AdVantage — unified SaaS for multi-channel ad campaigns with real-time analytics, lead tracking, and a cross-platform web + mobile dashboard.
• WebDevRef — offline HTML / CSS / JS docs PWA.

• TaskFlow — asynchronous scheduling engine for reliable task execution and micro-reminder distribution.
• PyNLP Toolkit — Jupyter notebooks for state-of-the-art text analysis and modeling.
• StreamAudio — yt-dlp Streamlit frontend for robust audio extraction.

• Edit > write. I prefer surgical changes to existing code over rewrites. Most repos here started small and stayed small on purpose.
• One source of truth per concern. OKF knowledge bundles, AGENTS.md files, and committed runbooks document architectural decisions inside the repo that owns them — not scattered across a wiki.
• Read the README, then the diff. Every project ships with a README that answers why does this exist, how do I run it, and what's the smallest thing I can change before getting into the code.
• Conventional commits. Every commit is a complete thought. No "wip", no "fix stuff".
• Donation-funded, not ad-funded. Buy Me A Coffee and GitHub Sponsors keep the lights on; nothing else is required.
• Locally reproducible CI. Dagger everywhere — dagger call ci on my laptop runs identical to CI. Portable across GH Actions, GitLab, Codeberg, self-hosted.

Daily drivers   

Also fluent   

Infrastructure    Cloudflare Pages + Workers + KV + D1 · Firebase Spark (Auth, Firestore, Storage) · GitHub Actions + Dagger · pnpm + MegaLinter · sops + age for env management · MIT throughout

AI / Agents    Claude Code + Claude Agent SDK · MCP servers (mine + community) · LangChain · RAG · Gemini API · LLM benchmarking

• Open-source AI / agent tooling — MCP servers, agent skills, eval harnesses, anything that makes Claude / GPT / Gemini agents more useful end-to-end.
• Cloudflare-edge architecture — anything that runs at the edge. I have strong opinions about Workers, Pages, KV, D1, and the right boundary between static and dynamic.
• Reusable CI patterns — Dagger modules, thin reusable workflows, umbrella-owns-secrets patterns for solo-dev polyrepos.
• India-data APIs — if you publish a free dataset India needs, I'll happily consume it and build a free public reader on top.
• Hire me? — open to roles that ship products and trust engineers. chirag127.in · LinkedIn

If you're building something in this space and want to compare notes, my DMs are open on every channel listed at the top.

_{Everything I publish is MIT. This profile lives at github.com/chirag127/chirag127.}