Skip to content
View Anmolnoor's full-sized avatar
🎯
Focusing
🎯
Focusing

Block or report Anmolnoor

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Anmolnoor/README.md

Anmol Noor

Software engineer building developer tools, full-stack products, and systems that make software safer to ship.

I like working where product engineering meets systems: turning ideas into useful software, understanding where it can fail, and improving the workflows around it. Lately, I’ve been building local-first coding agents and exploring how they can work independently while staying testable, governed, and under human control.

My work has also taken me through real-time applications, authentication, Web3, and self-hosted infrastructure. The common thread is curiosity, practical automation, and sharing what I learn as I build.

Portfolio Blog Book a call LinkedIn Anmol Noor Email Anmol Noor

What I’m working on now

  • skep: a local-first supervisor for AI coding agents, with sandboxed workers, independent re-verification, and human approval. It’s shipped: MIT-licensed, on PyPI (pip install skep), at v1.6.0, with docs at skep.anmolnoor.com. Recent releases added:
    • running skep from your phone as one conversation across Telegram, Discord, and Slack — including HIGH-risk approvals and /update;
    • governing ACP agents running inside other clients;
    • a policy you can browse like Group Policy, with a plain explanation for each rule;
    • schedules that can only narrow what their workers may do;
    • a Telegram alarm when an email carries an injection attempt;
    • research, YouTube, and a daily mail summary that work from your phone.
  • my-ai-ide: my IDE for the AI era. It isn't an editor: it's T3 Code running Claude Code and Codex side by side, plugins and skills that set how they plan, code, review, and ship, a lockfile that records the setup, and a weekly task that checks for drift and proposes upgrades. It also has the skills I wrote, with hard gates like no code before a failing test.
  • AuthWAPI: WorkOS-style self-hosted auth for localhost — one docker run, one portable data folder, hosted sign-in/sign-up pages, passkeys, 2FA, OAuth 2.1/OIDC, organizations, and a built-in admin MCP server (early development).
  • Foundation CLI: a shell-native coding agent built in public.
  • Beekeeper, agent-policy-gate, and agent-task-contract: ways to govern agent tools, review their work, keep useful audit trails, and hand tasks between a supervisor and disposable workers.
  • A homelab of 12 services on Proxmox, with Tailscale, shared storage, monitoring, and automation. Two laptops share the agent work and run tests in parallel. Write-ups: current setup, Tailscale remote access, and auto-updating containers with Watchtower.

Selected work

Project What it does
skep Supervises AI coding agents with sandboxing, verification, and approval gates.
my-ai-ide T3 Code + Claude Code + Codex setup with plugins, skills, lockfile, and weekly drift checks.
Foundation CLI A local-first, shell-native coding agent and engineering workflow experiment.
Beekeeper A governed runtime for multi-agent workflows.
agent-policy-gate Policy, review, and audit primitives for agent tool calls.
AuthWAPI WorkOS-style self-hosted auth for localhost with hosted pages, passkeys, 2FA, OAuth 2.1/OIDC, organizations, and an admin MCP server.
HoldIt A macOS menu bar utility for collecting files with drag and drop.
a2j A Chrome extension and local Python service that tailor resumes and cover letters to a job page.
skep-benchmarks A benchmark suite for LLM coding on realistic software engineering tasks.
claude-dotfiles and scripts A portable Claude Code setup and shell scripts for the same dev environment on every machine.

I’ve also built products including Amend Notes and Vendro, plus experiments in real-time software and Solana. The projects page has more, including older learning projects.

Open source contributions

Find your people

If you’re interested in local-first software, coding agents, agent safety, developer workflows, or homelabs, I’d be glad to compare notes or build together. I also enjoy talking about turning an early idea into a dependable product.

If you’re hiring for software engineering work across full-stack products, systems, or developer tools, the portfolio and selected projects are a good place to start. You can reach me through LinkedIn or email.

Writing

More on the blog.

Tools and technologies

TypeScript · React · Next.js · Astro · Node.js · Python · Rust · Swift · PostgreSQL · MongoDB · WebRTC · Solana · Docker · Proxmox · Tailscale · Prometheus · Grafana · Ollama · Claude Code


Snake animation

Profile visitors GitHub followers

Profile sources and editorial notes: PROFILE_CONTEXT.md

Pinned Loading

  1. scripts scripts Public

    Personal shell scripts & shortcuts for a consistent dev environment across machines.

    Shell 1

  2. hold-it hold-it Public

    Swift 1

  3. my-portfolio my-portfolio Public

    My portfolio webpage

    MDX 2

  4. claude-dotfiles claude-dotfiles Public

    Portable Claude Code setup — clone on a new machine, run ./install.sh, done.

    Shell 1

  5. beekeeper beekeeper Public

    Governed agent runtime with tool-level policy enforcement

    Python 1

  6. fcli fcli Public

    a local-first, shell-native coding agent.

    Python 1