iiinova — where your digital self keeps evolving

Vincent Behar @vbehar.dev · 17/09/2026

The C in MCP stands for Connectivity? ;-) yes, I'm at #mcpcon in Amsterdam this week

Reposted by Vincent Behar

Craig McLuckie @cmcluck.bsky.social · 01/07/2026

@h.olysh.it built an interesting OSS project: modelith.sh With AI driven development high levels of specificity upfront drives efficiency. 1. "grill-me" style skills to drive specificity 2. hybrid generated views (yaml + md w/ mermaid charts) 3. imperative linting forces correctness Super neat.

modelith.sh

Modelith - Domain Model Tooling | Modelith

Author, validate, and render domain models by talking to an AI agent.

142

Reposted by Vincent Behar

Marit van Dijk @maritvandijk.bsky.social · 29/05/2026

These would also be great for pulling Docker images on a conference wifi (ask me how I know)

A picture of plastic hands keeping a laptop lid open. The main text reads "Keep your agents running 24/7".

1153

Reposted by Vincent Behar

Brian Grant @bgrant0607.bsky.social · 26/05/2026

I created a Kubernetes package installer based on configuration as data. Instead of parameterizing every field, it encourages post-installation customization. The installer is open source and can be used without ConfigHub. Let me know what you think. medium.com/@bgrant0607/... #Kubernetes

medium.com

A new Kubernetes package installer

Introduces a Kubernetes package installer designed for simple installation and post-installation customization.

182

Vincent Behar @vbehar.dev · 26/05/2026

Nice! And congrats to Malo, I'll see soon if I prefer to work with the father or the son ;-)

020

Vincent Behar @vbehar.dev · 20/05/2026

Interesting: "It builds on top of Kubernetes features like Pods and Pod autoscaling, but takes the Kubernetes control-plane out of the critical path to achieve lower latency."

011

Vincent Behar @vbehar.dev · 14/05/2026

Cool. I'm using the MoE because I only have 12G. I wanted to wait for the official release but maybe I'll try the PR before that ;-) I'm curious to see if it makes the dense model fast enough to be usable.

000

Vincent Behar @vbehar.dev · 14/05/2026

Nice. What's the impact on ram and vram? As it's a 2nd model...

100

Vincent Behar @vbehar.dev · 14/05/2026

The links to parts 1 and 2 are wrong (404)

100

Vincent Behar @vbehar.dev · 14/05/2026

What ? You're not monitoring it with Prometheus / Victoria metrics ? ;-)

100

Vincent Behar @vbehar.dev · 14/05/2026

Hum... ERR_SSL_PROTOCOL_ERROR on your blog right now

200

Reposted by Vincent Behar

Michael Hausenblas @mhausenblas.info · 08/05/2026

Continuing my sabbatical builder spree: Meet otel.eexplorer, your friendly open-source OpenTelemetry Entity explorer. Feedback welcome here or via GitHub. github.com/mhausenblas/... #opentelemetry #entity #explorer #visualization #knowledgebase #sparql

091

Vincent Behar @vbehar.dev · 06/05/2026

I'm now addicted to 2 (small) pi extensions: github.com/jkuball/pi-d... and github.com/badlogic/pi-... It's way easier (and faster) to work with it when you can give it early feedback, in a nice editor. If you're using pi-coding-agent, check them out. If you're not (yet) using pi -> pi.dev

github.com

GitHub - jkuball/pi-discuss: inline-answers for your clanker from the cozyness of your editor

inline-answers for your clanker from the cozyness of your editor - jkuball/pi-discuss

030

Reposted by Vincent Behar

Andrew Nesbitt @andrewnez.bsky.social · 02/05/2026

A GitHub for maintainers - Giving dependencies the same treatment the fork got nesbitt.io/2026/05/02/a...

nesbitt.io

A GitHub for maintainers

Giving dependencies the same treatment the fork got

55118

Vincent Behar @vbehar.dev · 22/04/2026

I asked Claude (Sonnet) to bootstrap a new project (loosely) based on another one. There was a models.json file in the original project, with references to opus, sonnet, and haiku. Guess which ones made it to the new project? Only sonnet and haiku ;-)

010

Reposted by Vincent Behar

Coté @cote.io · 21/04/2026

Developers need to tinker or they'll reject your platform. You also can't let them customize everything - that's tomorrow's legacy code. Does it work to let them tinker at the edges, keep the center standardized? cote.io/2026/04/21/l...

cote.io

Let them tinker - hacking developer resistance to sound enterprise architecture and platforms

Developers need to tinker or they’ll reject your …

011

Vincent Behar @vbehar.dev · 17/04/2026

Excellent talk from @mariozechner.at on coding agents, and "slowing the fuck down" www.youtube.com/watch?v=Rjfb... 18 minutes, this is a very good time investment (while your agent is coding for you of course ;-) If you prefer to read: mariozechner.at/posts/2026-0...

youtube.com

Building pi in a World of Slop — Mario Zechner

YouTube video by AI Engineer

0153

Reposted by Vincent Behar

Adrian Mouat @adrianmouat.com · 15/04/2026

As someone who has suggested using cooldowns to mitigate some attacks, I did enjoy this article arguing they are a poor solution and moral problem: calpaterson.com/deps.html Cal suggests using upload queues instead, which I would agree with.

calpaterson.com

Dependency cooldowns turn you into a free-rider

Against dependency cooldowns as a response to supply chain attacks

201

Vincent Behar @vbehar.dev · 13/04/2026

"in PR descriptions, commit messages, and code comments [...] no emojis." What? But this was the killer feature of LLMs. Emojis everywhere ;-)

000

Vincent Behar @vbehar.dev · 13/04/2026

With a screenshot it's always better ;-)

110

Vincent Behar @vbehar.dev · 11/04/2026

If you're using the pi coding agent on Mac, here's a quick SwiftUI menu bar app to monitor your sessions: github.com/vbehar/PiScope and if you haven't tried pi yet, this is a good time to do it ;-) pi.dev

github.com

GitHub - vbehar/PiScope: A lightweight macOS menu bar app to monitor your [pi coding agent](https://pi.dev) sessions

A lightweight macOS menu bar app to monitor your [pi coding agent](https://pi.dev) sessions - vbehar/PiScope

121

Vincent Behar @vbehar.dev · 09/04/2026

Interesting (lighter) alternative to Temporal

021

Vincent Behar @vbehar.dev · 09/04/2026

Oh! pi.dev has moved to Earendil. Good move. If you are into agentic AI and haven't tried it yet, now is a good time.

pi.dev

A terminal-based coding agent

100

Reposted by Vincent Behar

Thomas Vitale ☀️ @thomasvitale.com · 02/04/2026

What if you could package, distribute, and secure Agent Skills like you do with your container images, configuration, and libraries? This is my proposal for an open specification for managing Skills as OCI Artifacts, re-using the same tools and infra you already have thomasvitale.com/agent-skills...

thomasvitale.com

Agent Skills as OCI Artifacts

I'm proposing a specification for packaging and distributing Agent Skills as OCI Artifacts for skills discovery, full portability, provenance, integrity, and enterprise features.

132

Reposted by Vincent Behar

Laurent Cheylus @lcheylus.bsky.social · 18/03/2026

Betterleaks: a new open source Secrets Scanner from the author of Gitleaks. It's a drop-in replacement for Gitleaks with new filters, configurable validation, faster scans, more options, and built for the agentic era. #Git www.aikido.dev/blog/betterl...

aikido.dev

Betterleaks: The Gitleaks Successor Built for Faster Secrets Scanning

Betterleaks is a new open source secrets scanner from the creator of Gitleaks. A drop-in replacement with faster scans, token efficiency detection, configurable validation, and more.

095

Reposted by Vincent Behar

Simon Aronsson @simme.dev · 17/03/2026

I just published a blog post on an experiment I did with building a dry-run mode for the OpenTelemetry Collector. It's a work in progress, and was mainly built to scratch my own itch, but nonetheless -- enjoy! ubuntu.com/blog/buildin... #observability #otel #o11y #sre

ubuntu.com

Building a dry-run mode for the OpenTelemetry Collector | Ubuntu

Teams continuously deploy programmable telemetry pipelines to production, without having access to a dry-run mode. At the same time, most organizations lack staging environments that resemble producti...

074

Reposted by Vincent Behar

Filippo Valsorda @filippo.abyssdomain.expert · 12/02/2026

The Go Checksum Database guarantees that every single Go build uses the same source for a given module version. However, GitHub might not show you that code. pkg.geomys.dev displays the canonical source of a Go module, and it comes with browser extensions to replace pkg.go.dev source links.

words.filippo.io

Inspecting the Source of Go Modules

Code hosts like GitHub don't necessarily show the correct source of Go modules. pkg.geomys.dev is a new convenient viewer for module source.

36913

Reposted by Vincent Behar

𝙳𝚘𝚝𝚊𝚗 𝙷𝚘𝚛𝚘𝚟𝚒𝚝𝚜 @horovits.bsky.social · 14/02/2026

interesting move by @percona.bsky.social, spinning off #opensource project PerconaEverest to a new co and turning it to a multi-vendor #database provisioning on #Kubernetes (OpenEverest). aiming to donate it to the CNCF. A nice addition to the #cloudnative stack. www.percona.com/blog/blog-po...

percona.com

Good Bye Percona Everest, Hello OpenEverest!

Percona Everest will become OpenEverest, retaining the same open source (Apache 2.0) license and become a multi-vendor project built with open governance at its core.

141

Reposted by Vincent Behar

Charity Majors @charity.wtf · 13/02/2026

o16g.com is the best thing i've read in a week full of seriously great reads. and now, as of two hours ago...with its own line of merch. do scroll down until you see the "honeycomb edition" shirts.. you will understand my excitement 😍🌈 www.bonfire.com/store/o16g/

bonfire.com

o16g | Outcome engineering merch | Bonfire

Check out o16g on Bonfire and shop official merchandise today! Featuring limited edition custom apparel, printed with care in the USA just for you.

1183

Vincent Behar @vbehar.dev · 12/02/2026

hum ça me dit que "Cet album a été consulté sur trop d'appareils" ;-)

100

Reposted by Vincent Behar

Adrian Mouat @adrianmouat.com · 06/02/2026

For those of us struggling with feeling devalued as programmers with the rise of AI, I present the following quote from Paul Graham: "In programming, as in many fields, the hard part isn't solving problems, but deciding what problems to solve."

131

Vincent Behar @vbehar.dev · 06/02/2026

Est ce qu'il y a un meilleur moment que le vendredi fin d'après-midi pour faire une release ? ;-)

000

Vincent Behar @vbehar.dev · 06/02/2026

Mitchell Hashimoto's AI adoption journey: mitchellh.com/writing/my-a... Really interesting. As always, push past the initial friction.

mitchellh.com

My AI Adoption Journey

010

Reposted by Vincent Behar

Fabrice Aneche @akhen.bsky.social · 05/02/2026

I really like the idea behind gryph github.com/safedep/gryph, it is hooking up onto your agents and create an audit trail log. It was simple to install, and did the job (I'm talking about the free opensource solution)

github.com

GitHub - safedep/gryph: The AI coding agent audit trail tool

The AI coding agent audit trail tool. Contribute to safedep/gryph development by creating an account on GitHub.

011

Vincent Behar @vbehar.dev · 05/02/2026

Merci à tous les orgas de la team @cloudnativedays.fr ! la barre était déjà haute après le KCD 2023, mais cette édition était vraiment top ! Vivement la prochaine ;-)

010

Reposted by Vincent Behar

Simon Willison @simonwillison.net · 30/01/2026

I wrote about Clawdbot/Moltbot/OpenClaw and Moltbook, the fascinating, weird and sometimes even useful social network for digital assistants to swap tips and gossip with each other simonwillison.net/2026/Jan/30/...

simonwillison.net

Moltbook is the most interesting place on the internet right now

The hottest project in AI right now is Clawdbot, renamed to Moltbot, renamed to OpenClaw. It’s an open source implementation of the digital personal assistant pattern, built by Peter Steinberger …

1724946

Reposted by Vincent Behar

Lachlan Kermode @ohrg.org · 28/01/2026

The Oxide RFD on using LLMs is exceptionally well written (h/t @bcantrill.bsky.social). This is a more coherent idea of what writing/text/language *is* and what's it's for---and why its expedited manufacture won't invalidate all aspects of that ontology---than a good chunk of the humanist hysterics.

rfd.shared.oxide.computer

576 - Using LLMs at Oxide / RFD / Oxide

0213

Vincent Behar @vbehar.dev · 09/01/2026

Wahou assassin! Ça nous rajeunit pas tout ça ;-) en tout cas bravo pour le reste, sauf les blessures, et bonne année à toi aussi. Faut qu'on se fasse un dej un de ces jours!

000

Reposted by Vincent Behar

Simon Willison @simonwillison.net · 31/12/2025

Here's my enormous round-up of everything we learned about LLMs in 2025 - the third in my annual series of reviews of the past twelve months simonwillison.net/2025/Dec/31/... This year it's divided into 26 sections! This is the table of contents:

The year of “reasoning”
The year of agents
The year of coding agents and Claude Code
The year of LLMs on the command-line
The year of YOLO and the Normalization of Deviance
The year of $200/month subscriptions
The year of top-ranked Chinese open weight models
The year of long tasks
The year of prompt-driven image editing
The year models won gold in academic competitions
The year that Llama lost its way
The year that OpenAI lost their lead
The year of Gemini
The year of pelicans riding bicycles
The year I built 110 tools
The year of the snitch!
The year of vibe coding
The (only?) year of MCP
The year of alarmingly AI-enabled browsers
The year of the lethal trifecta
The year of programming on my phone
The year of conformance suites
The year local models got good, but cloud models got even better
The year of slop
The year that data centers got extremely unpopular
My own words of the year
That’s a wrap for 2025

824954

Reposted by Vincent Behar

Bryan Liles @bryanl.dev · 28/12/2025

I have opinions about GenAI agents. Too many for Bluesky, so I wrote a whole blog post instead. blog.bryanl.dev/posts/agent-...

blog.bryanl.dev

Agents Done Right: A Framework Vision for 2026

Agents choke on context, loop on failures, and dump walls of code for review. It's time to rethink the architecture.

1295

Vincent Behar @vbehar.dev · 19/12/2025

Merci pour votre super travail !

100

Reposted by Vincent Behar

Sam Rose @samwho.dev · 16/12/2025

New post! ✨ Prompt caching ✨ My first big project post for @ngrok.com. 5 weeks, 12217 lines of code, 195 commits. I poured a lot into this one, and learned a lot in the process. I really hope you enjoy it ❤️

79520

Reposted by Vincent Behar

Simon Willison @simonwillison.net · 14/12/2025

Here's Emil's 17 step guide to how he used VS Code agent mode plus Claude 3.7 Sonnet, Gemini Pro 3 and Claude Opus to build the new library - it's a fantastic case study in using LLMs for serious, prediction quality code (vibe engineering, not vibe coding) friendlybit.com/python/writi...

friendlybit.com

How I wrote JustHTML using coding agents - Friendly Bit

I recently released JustHTML, a python-based HTML5 parser. It passes 100% of the html5lib test suite, has zero dependencies, and includes a CSS selector...

4474