How I Built and Open-Sourced 20+ AI Skills for CoderConsole 🚀

As developers, we lose an absurd amount of time to micro-context switching.

You are deep in a debugging session, and suddenly you need to decode a JWT, format a messy JSON payload, test a regex, or convert a timestamp. Usually, that means opening a new browser tab, searching Google, dodging three pop-up ads on a random utility site, pasting your data, and copying the result back.

Now that most of us work alongside AI coding agents in the terminal, that workflow feels even more broken. Why am I manually copy-pasting strings into a browser when my agent is right there—or worse, why is my LLM burning tokens hallucinating a one-off Python script just to base64-decode a string?

Over the weekend, I decided to fix this for my own workflow and built CoderConsole Skills—an open-source collection of 20+ everyday developer utilities designed to work seamlessly in both worlds:

1. As AI Agent / Terminal Skills (for CLI and agentic workflows)
2. As Clean Browser Utilities (for when you just want a fast visual GUI)

Install in Seconds (Codex, Claude Code & Antigravity)

You can install the entire skill pack (or specific skills) directly into your project terminal using a single curl command:

1. For Codex (--agent codex)

curl -fsSL [https://www.coderconsole.com/skills/install.sh](https://www.coderconsole.com/skills/install.sh) | sh -s -- --agent codex --scope project --skill all

2. For Claude Code (--agent claude)

curl -fsSL [https://www.coderconsole.com/skills/install.sh](https://www.coderconsole.com/skills/install.sh) | sh -s -- --agent claude --scope project --skill all

3. For Antigravity (--agent antigravity)

curl -fsSL [https://www.coderconsole.com/skills/install.sh](https://www.coderconsole.com/skills/install.sh) | sh -s -- --agent antigravity --scope project --skill all

The Core Idea: Deterministic Tools for Agents, Zero Friction for Humans

LLMs are great at reasoning, but they shouldn't waste context window or risk errors on deterministic utility tasks like hashing, encoding, diffing, or parsing. Giving an agent clean, reliable skills keeps it fast and accurate.

At the same time, some tasks are simply easier to inspect visually. Viewing a formatted JSON tree or tweaking a regex pattern with live highlighting is much faster in a browser UI than staring at raw terminal output.

CoderConsole Skills bridges both:

• Terminal / Agent first: Plug the skills directly into your AI agent setup so it can execute repetitive tasks locally without breaking your flow.
• Browser GUI: Use the exact same suite of tools on the web with zero setup or clutter.

What’s Inside (20+ Tools and Growing)

I started with the utilities I reach for every single week:

• Data & Formatting: [e.g., JSON Formatter & Validator, YAML to JSON, SQL Formatter]
• Encoding & Security: [e.g., JWT Decoder, Base64 Encoder/Decoder, Hash Generator]
• Web & Text Utilities: [e.g., Regex Tester, URL Parser, Diff Checker, Cron Expression Parser]

(Note: Swap the bracketed examples above with your actual top 6–8 tools before hitting publish!)

Open Source & Built for Contributions

I made the skills repository completely public so other developers can inspect how the tools are structured, fork them for their own setups, or add new utilities via Pull Requests.

If there is a repetitive task you hate doing manually—or a tool you wish your AI agent had out of the box—feel free to open an issue or submit a PR.

• GitHub Repository: github.com/nitiwari-dev/coderconsole-skills
• Agents Installation Guide: coderconsole.com/skills
• Live Browser Tools: coderconsole.com/tools

How do you handle dev utilities in your workflow?

I'm curious how others are approaching this in an agent-heavy workflow: do you rely on MCP/agent skills for deterministic tasks, let the LLM write scripts on the fly, or keep a browser tab pinned? Drop your setup in the comments!