Vibe Gain: How AI Unlocked Hidden Coding Time A year of GitHub data shows how AI collaboration transformed one developer's coding from sporadic bursts to sustained daily productivity—with 2-2.5x gains.

5/5: Full analysis: www.cyberchitta.cc/articles/os-...

Try vibe-gain: github.com/cyberchitta/...

Your AI coding boost? Reply below!

4/5: How? More sessions/day (1-2→3-4), faster iteration (gaps <5min: 38%→45%). Complex projects became doable.

All via chat AI (Claude/Grok + llm-context)—no IDE tools like Cursor.

#DevTools

Pre AI commit activity timeline

Recent AI commit activity timeline

3/5: Data: Active days +46% (99→145), commits/day doubled (3→6). Deeper work: repos with 100+ commits up (1→4).

See the shift:

#AIProductivity

2/5: Pre-AI: Bursty commits, big gaps. With AI: Steady output, 2-2.5x median LOC/day (225→532).

AI made any time slot productive—brief or long, switching tasks felt instant.

#AIDev #OpenSource

🧵 1/5: 🚀 New post: "Vibe Gain: How AI Unlocked Hidden Coding Time"

GitHub data shows my coding shift from sporadic bursts to daily flow—2-2.5x gains via AI. No extra hours, just frictionless sessions.

Thread: Metrics, charts, insights 👇

#AI #Coding #Productivity

I added evals for gemini-2.5-pro-exp-03-25 and deepseek-chat-v3-0324. tldr; gemini-2.5 one-shotted it. deepseek-3 two-shotted it.

bottom line. i think my eval has been saturated now. I likely will not test any more models on it.

GitHub - cyberchitta/scrapling-fetch-mcp: Simple URL fetcher MCP server for Scrapling. Retrieves HTML/markdown from automation-blocking sites with three protection levels. Simple URL fetcher MCP server for Scrapling. Retrieves HTML/markdown from automation-blocking sites with three protection levels. - cyberchitta/scrapling-fetch-mcp

So I built scrapling-fetch-mcp! A simple URL fetcher that helps LLMs access websites protected by bot-detection.

GitHub: github.com/cyberchitta/...

Felt weird that Claude couldn't fetch a URL (using mcp-fetch) that I had open in a browser window right next to it. Even weirder that I found an MCP tool for this purpose, but needed to get an API key to use it.

I just added evals for grok-3-beta and grok-3-beta-think. grok-3-beta-think hit the top-tier, grok-3-beta not so much.

Added evals for o3-mini and o3-mini-high

Added evaluation for gemini-2.0-flash-thinking-exp-01-21

Added evaluation for deepseek-r1

Added evaluation for minimax-text-01

Added evaluation for deepseek-v3

Added evaluation for gemini-exp-1206

Added evaluation for gemini-2.0-flash-exp

28 Alternatives to LLM Context A comprehensive list of open source tools that help developers pack their code into LLM chats. From simple file combiners to complex context managers, these CLI tools streamline the way you share code...

📝 28 Alternatives to LLM Context

We found 28 different tools for sharing code with LLMs - and built one more! Check out our roundup of open-source alternatives to LLM Context, and see how developers everywhere are tackling this challenge: www.cyberchitta.cc/articles/lc-...

Added evaluation for QwQ-32b-preview

Added evaluations for llama-3.3-70b-instruct and nova-pro-1.0

Added evaluation for o1

GitHub - cyberchitta/llm-context.py: A command-line tool for copying code context to clipboard for use in LLM chats A command-line tool for copying code context to clipboard for use in LLM chats - cyberchitta/llm-context.py

📢 llm-context v0.2.0 is out!

✨ Major feature: Full Claude Desktop integration via MCP
- Seamlessly explore codebases right in your conversations
- Profile-based content filtering
- Intelligent context selection

🔗 github.com/cyberchitta/llm-context.py
📦 uv tool install llm-context

LLM Context: Harnessing Vanilla AI Chats for Development The case for a tool that enables efficient use of web-based AI chat interfaces for software development, offering an alternative to IDE-integrated solutions.

🚀 Introducing LLM Context - a tool to seamlessly integrate AI chat into your coding workflow. Discover how it enhances development through efficient context management and workflow integration. Check it out:

🔗 www.cyberchitta.cc/articles/llm...

Added evaluations for #qwen-2.5-coder-32b-instruct and #gemini-exp-1121

I just added evaluations for #qwen-2.5-72b-instruct and #nemotron-70b-instruct

14 LLMs Tackle Shopify: Unexpected Results What happens when 14 LLMs—ranging from industry giants to up-and-coming contenders—tackle a real-world Shopify optimization problem? Find out which models delivered and which stumbled in this in-depth...

Smaller LLMs beat GPT-4 at Shopify code refactoring! 🤯
Claude-Haiku & Mistral outperformed industry giants in our 14-model comparison.
Full analysis: cyberchitta.cc/articles/llm...

#deepseek-r1-lite-preview added