rohitghumare

Looks very promising.

I also wrote more about this: https://rohitghumare.com/blog/stateless-mcp/

#1 Persistent memory for AI coding agents based on real-world benchmarks.

Persistent memory for Claude Code, GitHub Copilot CLI, Cursor, Gemini CLI, Codex CLI, Hermes, OpenClaw, pi, OpenCode, and any MCP client.

I’ve been using the new SkillKit CLI for myself as a universal skills layer on top of all the agents I use (ChatGPT, Codex, Claude Code, Cursor, Copilot, Windsurf, Devin, etc.). Instead of rewriting the same skills per agent format (SKILL.md, .mdc, .github/skills, etc.), I write once and let SkillKit install, translate and sync to 45+ agents automatically.

The SkillKit vs skills CLI video I recorded shows side‑by‑side comparisons: download speed, sparse checkout on large monorepos (~1.4s), security scanning, and how it bulk‑removes skills per repo/source. It also covers how SkillKit keeps everything portable and secure across registries.

Repo: https://github.com/rohitg00/skillkit · Website/docs: https://skillkit.sh

https://Skillkit.sh already does this by default and ranked #3 on Product Hunt.

If you really looking for memory solution for your agents: https://github.com/rohitg00/aegntmemory

If you'd asked me last year to run an autonomous research loop across two GPUs, I'd have said that's not something I can do.

Not "it'll take a while." Impossible.

Today this is my Saturday.

My - - : autonomous hyperparameter mutation, parallel experiments, no babysitting.

Results: → 17 experiments, 0 crashes → baseline 1.2365 → . _ → 1.48% improvement. Found by the loop.

The chart shows the staircase, the same pattern Karpathy sees in his runs. His: 2 days, 276 experiments. Mine: 1 hour, 17. Same logic, different constraints.

The constraint here is the ' _=, which means ~5.5% MFU. The GPU is mostly waiting on memory transfers, not computing.

What the same loop looks like on different hardware: ( ): ~ / → .% : ~+ / → -% (): ~+ / → -% (): ~+ / → -%

Same autoresearch loop. Just more runway.

Karpathy's setup, 8x H100 running 48 hours would likely hit 5,000+ experiments. On 4090s that's not feasible. But 24-48 hours on what I have would still find significantly more than 1 hour did. That's what's running next.

This is my own multi-GPU implementation built on top of his single-GPU original, orchestrated with iii functions, workers, and triggers.

Claude Code made the extension possible in a weekend.

If you want to try: https://github.com/iii-hq/n-autoresearch open source repo.

Built n-autoresearch - As I saw, in comments on X, there is a need to run agent swarms in parallel on multiple GPUs, hence adding a structured continuation to the autoresearch project by legend Karpathy's, powered by Worker / Function / Trigger as primitives.

What's different: → 21 functions, 23 triggers, 8 KV scopes for structured experiment state → Polyglot: TypeScript orchestrator + Rust GPU worker → Multi-GPU parallel experiments with adaptive search (explore → exploit → combine → ablation) → External agents call functions via REST: no LLM baked in, similar to autoresearch itself.

Same val_bpb hill-climbing loop, but with proper state management, crash recovery, near-miss tracking, and structured reporting.

Testing on NVIDIA GB10 (Grace Blackwell).

By adding pure iii-sdk worker for json-render UI generation with JSONL patch streaming, caching, rate limiting, and validation. No standalone HTTP server, everything is just endpoints that are iii functions with HTTP triggers served by the iii engine.

1.1K stars

It brings agent swarms aka teams to claude code with this: https://github.com/rohitg00/pro-workflow

But it takes lot of context as a experimental feature.

Use self-learning loop with hooks and claude.md to preserve memory.

I have shared plugin above of my setup. Try it.

is by far the most amazing thing that happened in 2026

That's why https://agenstskills.com validate every skills

Day 9 dropped already

Motia brings cohesion to the fragmented backend world with our core primitive: the Step.

Something I was looking for.

Here’s the core document system:

Product Requirements Document (PRD)

App Flow Document

Tech Stack Document

Frontend

Backend Structure Doc

Implementation Plan

*Project Status

It's good, I tried it.

Live demo: https://youtu.be/uyo3dfR9DDc?si=0aVuqmRm-Libc1Sl

In this video we can learn about: A quick overview of the Redis license change. An introduction to Valkey and its goals. A hands-on performance benchmark comparing Redis and Valkey using official Docker images. We test key operations like SET, GET, LPUSH, LPOP across different data sizes. Analysis of the benchmark results to see if Valkey truly performs as a drop-in replacement. Mention of managed Valkey options like AWS ElastiCache for Valkey.

Have you ever struggled with sharing Full Stack Application components across projects? Or found yourself copying and pasting code repeatedly? If you've said yes to this, you're not alone. These are exactly the problems that led me to explore Bit.

Bit is a platform for building applications using independent, reusable components. Unlike traditional approaches where components are secondary to your application, Bit puts components at the center of development. This creates a more modular, maintainable, and collaborative way to build software.

What kind of discussion? I've tried it, let me know if I can help here

I found out about Harmony recently from their founder, they say it's an open source stitching layer for composable architectures.

1. An AI-Powered Mathematical Animation Generator

2. Company-Wide Self-Hosted File Sharing Solution for Teams

3. Your Own GIF Hosting Platform

4. AI Design Team using Gemini Flash 2.0

5. Host Your First AI App in Seconds

6. Build an AI Math Animation Generator

7. Three tech words

Agree, but doesn't it will increase the space on clean website

Projects: - https://www.devopscommunity.in/

- https://interview.devopscommunity.in/

- https://tools.devopscommunity.in

- https://whopayswriters.me/

- https://ai-web-summarizer-two.vercel.app

- https://github.com/rohitg00/host-your-gif

- https://github.com/rohitg00/manim-video-generator

- https://github.com/rohitg00/ai-food-recipe-assistant

- https://github.com/rohitg00/ai_computer_use

- https://github.com/rohitg00/process-viewer

- https://github.com/rohitg00/my-own-whiteboard

- https://github.com/rohitg00/s3-file-share-for-free

- https://thatdevopsguy.substack.com/

Yes, feel free to try open-source version available on Github.

Just published a new release to support multiple cloud storage options.

Thanks

New Release is out now

> We support 7 cloud providers now. > Beta storage providers are still being tested.

Stable: Amazon S3 ($23/TB/month) Google Cloud Storage ($20/TB/month) Cloudflare R2 ($15/TB/month)

Beta/Testing: Wasabi ($6.99/TB/month) Backblaze B2 ($6/TB/month) DigitalOcean Spaces ($5/TB/month) Hetzner Storage ($4/TB/month)

Never came across this, Interesting project. Thanks for sharing

Quick Update: I didn't expect so many people to show interest in this project. As a few suggested, I plan to add more storage solutions in the next few days.

Testing Other Cloud Providers:

Backblaze B2 ($5/TB/month)

Wasabi ($6/TB/month)

Google Cloud Storage ($20/TB/month) - I am ready to deploy to production.

DigitalOcean Spaces ($5/TB/month)

Cloudflare R2 ($15/TB/month)

Hetzner Storage ($3/TB/month)