DEV Community: adriens

📊 I (finally) built the demo of my duckdb brew extension

adriens — Fri, 29 May 2026 08:07:18 +0000

This is a submission for the GitHub Finish-Up-A-Thon Challenge

💭 Where it comes from

In the previous GitHub Copilot CLI Challenge I submitted a realization ; a duckdb extension to make brew packages reporting easier for analysis and security/audit concerns :

GitHub Copilot CLI Challenge Submission

adriens

Feb 14

🦆Built a duckdb community extension for brew

#devchallenge #githubchallenge #cli #githubcopilot

3 min read

Since that my extension got accepted as part of community extensions (see dedicated page), I maintain it along duckdb releases and have an constant 824 downloads/month, slowly gaining in popularity :

// Detect dark theme var iframe = document.getElementById('tweet-2058293935114801558-705'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=2058293935114801558&theme=dark" }

I've provided a set of ready to use SQL, yet for some it may not show the full potentiel that I had in my head : I needed to make it more visual with dataviz and professionan datascience tools.

I must confess I was a bit procrastinating that part... But this DEV Challenge PROMPT triggered the action!

🤗 What I Built

I have created a Python based Quarto Notebook that generates a nice looking pdf report, thanks to the extension, ad put Graph Datascience in it.

🍿 Demo

🤔 The Comeback Story

Images are much better than words :

// Detect dark theme var iframe = document.getElementById('tweet-2060466960794853574-236'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=2060466960794853574&theme=dark" }

🦥 My Experience with GitHub Copilot

I built the Quarto notebook within 2 hours of prompting before getting out of tokens :

⚚ I reverse-engineered my mobile operator's APK — then Hermes Agent wrote the executive report

adriens — Thu, 28 May 2026 23:01:45 +0000

This is a submission for the Hermes Agent Challenge: Write About Hermes Agent

// Detect dark theme var iframe = document.getElementById('tweet-2059698327235805258-575'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=2059698327235805258&theme=dark" }

🍿 Demo

📸 Screenshots

Just check how amazing these reports look like!

Building a Multi-Stakeholder API Monitoring Report with Hermes Agent

"Just generate a PDF" — famous last words. What started as a simple request turned into something much bigger: a full monitoring stack for my mobile operator, built evenings and weekends, culminating in a 17-page professional report that would have taken a full week to build manually. Total cost with Hermes Agent: $19.57.

🏝️ The Backstory: A Stack No One Else Has

I'm Adrien, a developer in New Caledonia. OPT-NC's Helia mobile service has an app — but no public API, no CLI, nothing on any marketplace. So I reverse-engineered the APK, extracted the private HTTP calls, and rebuilt them in a Go CLI that snapshots voice, data, and SMS consumption every 5 minutes into a local DuckDB database.

Then I built KDE Plasma widgets in Python/PyQt that read from DuckDB and display live on my desktop — mirroring the official app's data, but with history, trends, burn rate, and alerts. Plus a system tray icon showing live API status at a glance.

The official app shows you now. My stack shows you now, history, trends, and alerts.

No other Helia customer has this. That's the breakthrough.

🎯 The Problem Hermes Agent Solved

All this data was sitting in DuckDB. The question was: how do I present it to people who don't speak SQL?

Three completely different audiences: the CEO who needs a 30-second summary screenshot-ready for PowerPoint, the CIO who wants ROI in euros and SLA compliance, and the Network Admin who needs actionable tickets with specific hours and error patterns.

🎭 The Role-Playing Game That Became the Design Document

Before writing a single line of code, I asked Hermes Agent to do something unusual: simulate a full team meeting with 7 personas — CEO, CIO, network admin, developers, and marketing.

It produced a full transcript. Each persona argued their case:

The CIO: "I need ROI in euros, not percentages."
The CEO: "I need the 30-second version I can project tomorrow morning."
The Network Admin: "Give me Jira tickets with specific hours, not a dashboard."

That transcript became the design document. Every page of the report was written against a specific person's stated need. No guessing. No generic output.

⚡ What Hermes Agent Built in ~1 Hour

It started with the data. No assumptions — it queried the schema and immediately caught something: average latency was 2,534ms but the median was 204ms. Bimodal distribution. That single insight shaped every chart.

It extracted brand colors from the website before writing a single line of LaTeX. Hermes Agent opened Helia's site, pulled the magenta/pink gradient from the SVG logo, and used it consistently across every chart, table, and tcolorbox. Small detail. Big difference.

Then it built everything:

4 Python scripts: latency distributions, timeout heatmap, sparkline, 4-panel executive chart
Full XeLaTeX report with TikZ progress bars and brand-colored boxes
A French accent fixer script baked into the pipeline

One message: "update with fresh data" — triggered 8+ tool calls automatically: DuckDB → diff → scripts → charts → LaTeX → 2× compile → verify.

📊 The Charts It Produced (Without Being Asked)

I asked for charts. I didn't ask for this:

Latency distribution — median/mean/P95 lines labeled, shaded fast vs slow zones, annotated arrow pointing to the long tail: "Queue longue (timeout ~12s) (~40% des pings)". Log-scale version revealing the true bimodal structure: two peaks at ~80ms and ~4s.

4-panel executive dashboard — availability gauge (87.2% vs 99.5% SLA), latency (2534ms vs 500ms — "5.1x trop lent"), timeout rate ("1 requête sur 8 échoue"), composite SLA score per metric. Score global: 65%. Verdict: RED.

Timeout heatmap — all 18 timeouts concentrated on Wednesday evening 19h–22h. The rest of the week: clean. Instant actionable insight for the network admin.

All in Helia brand colors. All annotated. None of it explicitly requested.

📋 The Executive Summary: Fits in One Slide

Page 3 of the report is designed to be screenshotted directly into PowerPoint. Verdict box at the top (red), plain-language translation ("1 call in 8 fails"), business impact in euros, top 3 problems with responsible parties. The CEO gets it in 30 seconds.

🧠 What I Learned

Orchestration is the real superpower. I asked for a PDF. It fetched brand colors, queried the DB, wrote Python scripts, compiled LaTeX — all unprompted.

Iteration is the default. The first charts were basic. The final ones have annotated thresholds, color-coded data points, and statistical summaries. That's the loop.

Skills are the real ROI. Everything got distilled into a reusable ~/.claude/skills/reporting-latex/ skill. Next similar project starts at 80%, not zero.

💰 The Cost

Qwen3.7-Max via OpenRouter. Alibaba's flagship agentic model, built for long-horizon autonomous execution.

326 requests. 61.4M tokens. $19.57. ~1 hour. vs a full week manually.

🖥️ What's Next

The next step: run it all locally. I'm currently eyeing an Apple Mac Studio M4 Max — 16-core CPU, 40-core GPU, 64 Go unified memory. New Caledonia is 20,000km from everything; local inference just makes sense. Zero API costs, zero latency to the cloud, full control.

With 64GB unified memory, I'd be able to run:

Qwen3-Coder-Next — purpose-built for agentic coding, 70%+ on SWE-bench, needs 46GB RAM ✅
Qwen3.5-35B-A3B — the 2026 community default for local inference, ~80 tok/s via MLX ✅
Qwen3.6-27B — optimized specifically for agentic coding workflows ✅

The same class of model as Qwen3.7-Max on OpenRouter — locally, for free, forever. A fully local Hermes Agent stack.

From $19.57 on OpenRouter to owning the hardware. That's the roadmap.

The pattern: monitor → query → visualize → compose → translate. The translate step — turning a percentile into a story a CEO can act on — is where Hermes Agent earns its keep.

Have you used Hermes Agent for multi-tool orchestration? Curious how your experience compares.

🧞‍♂️Transform unstructured PDFs Job Offers into a dataset w. gemma4:2b

adriens — Sun, 17 May 2026 22:47:08 +0000

This is a submission for the Gemma 4 Challenge: Build with Gemma 4

🤔 About the power of collections and our ability to compare things

First a bit of philosophy.

Did you notice how we tend to align things, tend to shape things so they can be aligned, compared (based on a common attributes like color, weight,...).

Comparing objects is much easier when they share common structure, then we can use attributes to get more knowledge, produce KPIs, make clever choices and put things in evidence.

👉 Well the same happens with machines : it's much much easier to compare and manage documents when they share a same structure.

This is the very core idea that motivated this work to explore how Open Source AI could help in a very pragmatic way... and feel the opportunities it opens with concrete prototypes.

🙋 What I Built

I've built a whole real life and live data pipeline that takes as input Open Data Public Sector Job offers (dataset/avis-de-vacances-de-poste-avp-drhfpnc) :

csv
Raw PDFs

Then,

From a dedicated GH repo adriens/avps I've prepared a whole structured mix of md thanks to csv
Then from the GH Action I did transform brut raw PDFs with pypi.org/marker-pdf into markdown
I ended to publish a dedicated Zensical gh-pages website : adriens.github.io/avps

Next, this is where things go really interesting : I wanted to be able to compare job offers the one against the others... but markdown were far too much different the one from others :

Not the same number of sections
Not the same section titles : hard skills, soft skills, missions,...
Not the same levels of sections
Not necessarly itemized the one
Not the same style at all (section levels, CAPITALs, email, cities...)

... which made it very hard... or even impossible to compare them amongs the others... or even crazier : put them in a traditional SQL structured database.

👉 This is where gemma4:2b comes in to create a very well and consistent set of markdowns that can then be used for various use cases :

Create very well structured ePub to read job offers on the go (and docx)
Create a very clean and well organized PDF : very easy to load in assistants, print or to drop in any assistant
Deliver structured data with clean json files
Make a duckdb database and perform SQL on the data by using the now well structured markdowns, which made it possible to open unprecedented and exciting reporting opportunities (here in duckdb)
Share all this as a dataset on Kaggle

SELECT '--- RÉPARTITION DES COMPÉTENCES PAR DOMAINE ---'
as titre_report;
SELECT domaine, count(*) as nb_competences 
FROM savoir_faire 
GROUP BY 1 ORDER BY 2 DESC;

// Detect dark theme var iframe = document.getElementById('tweet-2055533578021396775-283'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=2055533578021396775&theme=dark" }

🎯 Problems it solves

In input we really had very various kind of PDF documents, and no structured tabular data, now, they both are delivered as :

Well formated and structured markdown
A real database that embeds data as tables and views for advanced SQL reporting and charting
Ready to use and perfectly well structred ePub and PDF documents, very easy for LLMs to understand

🤗 Experience it creates

The experience is rather an data experience as thanks to data normalisation and standardization we can load and compare job offers, which make job search and indexing much much more efficient, whatever the input.

Last but not least, using gemma4:2b-it proves that great things can be achieved even with small resources and that well prepared data opens so many intelligence opportunities, without having to deal with frontier models as "the output I got is good enough".

🍿 Demo

💰 The benefits : then and now

Below the benchmark of markdown before and after

⚖️ Benchmark : `marker-pdf` vs. `marker-pdf` ➕ `gemma4:e2b`

Below some results:

Structure consistency:

📊 Analytics on top of database

One the well-structured json could be produced from the markdown I could efficiently load them into a duckdb database and do some reporting see AVPS DRHFPNC - Les pdf en SQL avec duckdb Kaggle notebook :

// Detect dark theme var iframe = document.getElementById('tweet-2055751934301380738-571'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=2055751934301380738&theme=dark" }

📜 Code

Kaggle Notebook : IA AVPs DRHFPNC Structurés
Kaggle dataset : avps-nouvelle-caldonie-structurs/data
GitHub repo adriens/avps
Zensical website : adriens.github.io/avps
Kaggle notebook that shows how to load structured json into a duckdb database : AVPS DRHFPNC - Les pdf en SQL avec duckdb

🎁 Goodies

Notebook LM

💡 How I Used Gemma 4

I chose google/gemma-4/transformers/gemma-4-e2b-it from kagglehub as I had a huge amounf of data to load (all New-Caledonia ones) and a restricted amount of time on Kaggle as well as small GPUs.

Also my intent was to be able to run this code one day onPrem on my very own hardware so I decided to stay as little as possible.

🤔 What remains to do...

Try to:

Add an evaluation phasis to check output consistency
Try to switch to CPU mode so the Notebook can be scheduled without exceeding the maximum Kaggle window
Use gemma-4-E4B and benchmark output quality
Produce native adoc with proper annotations
First produce json (and more standardized values, enums,...) then re-generate md/adoc from it

💸 How $0 marketing budget boosted my creativity & automation opportunities

adriens — Wed, 11 Mar 2026 20:11:05 +0000

💡 Inception back in days

Years ago I watched the following Netflix series :

A look beyond blueprints and computers into the art and science of design, showcasing great designers from every discipline whose work shapes our world.

Then wrote a dedicated article :

How did Netflix put my heart & brain on fire with "Abstract: The Art of Design" S02

That brought me a lot of inspiration about :

The power of simple/pure designs
Importance of typography
How to manage color gradients
How a minimalist design opens opportunities
How playing with things makes us creative and productive (Cas Holman, founder of toy company)

⏳ About `geol` : the need for a logo

Then since last year we started to work on geol, and I started to feel the need to get a logo for example to :

Display in the terminal
Show up in youtube thumbnails
Give an identity to our website (we built one w. Docusaurus)

So I filled an issue... to fix later.

🎨 How low marketing skills and budget boosted my creativity

// Detect dark theme var iframe = document.getElementById('tweet-1992068608395276623-107'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=1992068608395276623&theme=dark" }

🍿 Show me what it looks like

📝 Inception and paper/pen drafts

To build a logo, you need to focus on what your product does at its very core.
In my case, I came to the conclusion that there were two core things:

⏳ Pay a tribute to the data source : endoflife.date API
🚦 Colors which is the color identities, know when:
- 🟢 We're ok with a product,
- 🟠 When we should start to take care: time to plan
- 🔴 When it's too late: you're at risk

So finally, two dimensions :

📐 Shape
🎨 Colors

It started to trigger my curiosity about geometric considerations, which could hence be described thanks to mathematical objects, finally opening coding and automation opportunities.

I started to draw and draw a lot of shapes, think about how to put colors, the benefits of a simple geometrical/mathematical vectorial approach while enjoying a coffee cup outdoor seating at a café, drafting ideas and charts w. the paper/pen strategy :

// Detect dark theme var iframe = document.getElementById('tweet-2025338147153084787-440'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=2025338147153084787&theme=dark" }

☝️ Put more constraints to be more productive

I really love the pattern that consists of putting more rules to be more creative and just:

see what happens if...

Finally I got my own rules.

The logo should:

Be perfectly described with a few mathematical geometrical rules as possible (easier to code or to draw, even by a kid)
Give sense to colors and product lifecycle
Pay a tribute to its core meaning : the hourglass
Fun with graph datascience: it has to be both eulerian and hamiltoninan
Be as light as possible

🚀 Pure code and automation

Once I found the solution came the coding part:

Design the svg (and put placeholders inside them)
Create a Taskfile to automate a bunch of things (generate various formats from a single task command)
Inject colors with code and automation
Create automation tasks that build jpeg, png, animated & static gifs, webp, optimize jpeg... all built and delivered as output artefacts

So finally I came from a workflow on a paper to a fully automated makefile:

// Detect dark theme var iframe = document.getElementById('tweet-2030389652960710939-827'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=2030389652960710939&theme=dark" }

🖼️ Benefits : the `100%` LaTeX cheatsheet

As a benefit I could produce a 100% LateX version of the logo (see full demo)

// Detect dark theme var iframe = document.getElementById('tweet-1995616096351891548-579'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=1995616096351891548&theme=dark" }

🔭 Further opportunities

What's exciting with a "simple logo" is that it makes it possible to use the mathematical vectorial description and inject it in many artistic artefacts

3d printing
Pixelart
T-Shirts
Code driven animations
Shaders
Desktop Wallpapers

Who knows where creativy will take us 🤗

🦈 Analyzing Open Data shark attacks with gemini-cli and Quarto 📊

adriens — Sat, 28 Feb 2026 06:55:37 +0000

This is a submission for the Built with Google Gemini: Writing Challenge

😂 Spoiler alert : what it was not

It was not a straight and peaceful journey

// Detect dark theme var iframe = document.getElementById('tweet-2027947601442836653-336'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=2027947601442836653&theme=dark" }

❔ What I Built with Google Gemini

I created a Quarto Notebook based on Open data (I maintain since many years now) that reports Shark attacks in New-Caledonia to better understand what happens since many years now.

I used Gemini-cli to :

Help implement the Notebook that I wanted to look a bit like a newspaper report
Give me some inspirations during chats
Ask him to challenge me with questions
Imagine an interview which gave me the opportunity to better express & share my ideas
Challenge me with new questions
Finally produce a journalistic like paper anyone can read and print
Finally put this pdf as input with other related articles data sources into NotebookLM to produce a video

Pretty fast, I could get first stuning dataviz :

// Detect dark theme var iframe = document.getElementById('tweet-2027632648970223954-198'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=2027632648970223954&theme=dark" }

🔥 Work planning (brain on fire)

Saturday 2026-02-28 : 07:00 AM to 18:30 PM
- First MVP
- Video demo
- NotebookLM video production
- First article version and publication
- First social posts to share the insights I discovered
Sunday 2026-03-01 07h00 AM - 10:30 & 13:00 - 15:00 :
- pdf finetuning, review and little enhancements
- Section title review
- Charts title review
- Take feedbacks and implement them
- Chase on keyfindings hallucinations with specific data-driven PROMPTs

🍿 Demos

The main output of the project is a pdf.

Yet, here is the full storytelling I'll use for educational purpose:

Long edition with pdf commenting

Short edition (for educational purposes)

📑 Resources

Source code : github.com/adriens/odata-ncl/data/shark-attacks
Final pdf report
NotebookLM project
Python SDK for water quality : adriens/edb-noumea
Ready to use csv data for water quality : edb-noumea-data, automated with Gh Actions
Go TUI to browser water quality from any terminal : edb-noumea-tui
Kaggle dataset : 🦈 Shark Attacks - New Caledonia
Kaggle Notebook : 🦈 Shark Attacks New-Caledonia for dummies
Kaggle Notebook : Shark Attacks in New Caledonia - Analysis
Kaggle Notebook : 🦈 Shark Attacks New-Caledonia KPIs

🧑‍🎓 What I Learned

Need to very seriouly & carefully review AI generated comment on charts as quite often for example the "Key findings" were wrong :
- Nor respect for the color code accross the whole document (fatal vs. non fatal attacks)
- Evaluating 59% as the 2/3 of the attacks
Put the terminal to a given level of transparency so I coul read the resulting pdf while chating with gemini-cli (I have a small laptop with single screen)
I could very efficently take other people feedbacks in account which had a huge impact on dataviz and KPIs to avoid bias

💡 Google Gemini Feedback

I only used Gemin-cli :

I really enjoyed being interviewed by Gemini, it really made me creative, it was the first time I did this, but for sure I will reuse this pattern

📸 Screenshots

Below a collection of the resulting pdf :

🦆Built a duckdb community extension for brew

adriens — Sat, 14 Feb 2026 03:55:40 +0000

This is a submission for the GitHub Copilot CLI Challenge

🤓 What I Built

Sometimes ago I got asked to report what I installed on my Ubuntu... and like many people it turned out I installed various softwares in very various ways :

apt
snap
brew
pip or uv (but mostly in virtual environments)

Basically, at first I started with that kind of data-analysis and dataviz (and made it open source, see adriens/ubuntu-rmarkdown-report):

So I started to look around how I could efficiently automate that with standard data tools to make this work look like a piece of cake (avoid boilerplate R code in my case).

I then found the great osquery initiative, which makes it possible to query any operating system metric or information from a simple SQL :

osquery exposes an operating system as a high-performance relational database. This allows you to write SQL queries to explore operating system data. With osquery, SQL tables represent abstract concepts such as running processes, loaded kernel modules, open network connections, browser plugins, hardware events or file hashes.

But when I wanted to query installed brew packages, it turned out that I had to write an extension (which is very well documented).

So I finally opted to write a dedicated Rmarkdown notebook, started to play with custom dataframes (easy to share)... but pretty soon I felt like it would be much more convenient to be able to query this data within duckdb itself so it could make the data much much easier to get in various usecases (R, Python, Go, ... ).

💡 The idea : the `duckdb` extensions community

Therefore, I decided to build - and heopefully to get it accepted - a duckdb brew community extension so I can build reports on the installed homebrew packages from simple SQL and benefit from the amazing duckdb ecosystem and a very large amount of language bindings and data tools:

// Detect dark theme var iframe = document.getElementById('tweet-2016990414583321074-741'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=2016990414583321074&theme=dark" }

🍿 Demo

Official duckdb's brew DuckDB CE extension's page.

Below the whole story, from the function table concept in various databases, to the pitch, including the developer experience with Copilot CLI:

📷 Screenshots

📈 Historical brew install

When doing reports, I felt the need to know how many brew packages I installed, and when:

🌌 Graph Dependencies

For security analysis, it makes sense to be aware of the centrality of the installed software.

It is now made easy to produce and chart, see below:

🐧 `SQL` reports from terminal

Then it's really very easy to do reports from the terminal itself:

🤓 My Experience with GitHub Copilot CLI

Below my complete development workflow setup, which consists in three panels :

The top left one : to run make commands by myself, which made things easier to handle
The bottom left one : the GH Copilot CLI to which I asked to add features, write nice commit messages, check CI for failures, write Release Notes, create tags and releases
The right one : dedicated to try the output binary by myself, purely manual written SQL statements etc...

This how it looks::

Video announce :

// Detect dark theme var iframe = document.getElementById('tweet-2019167438546563160-116'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=2019167438546563160&theme=dark" }

👨‍🎓 Lesson learned : `CI`, tests and lint first

The CLI experience is very responsive, but what I would keep in mind and would like to put the maximum emphasis is how crucial are about build automation and CI :

Tests
Linting process
Code formating rules

The more they are part of the build tools and CI pipeline, the more the vibe coding experience is efficent and the more we build trust with the generated code.

In my case, duckdb community provided a repo template (see duckdb/extension-template) with all the required stuff so I could very efficiently start to work, which means : get continuous feedback from CI and give it immediately to Copilot for continuous improvement without having to ask help to community.

📑 Resources

Extension source code : adriens/duckdb-brew
duckdb brew extension homepage : https://duckdb.org/community_extensions/extensions/brew
adriens/ubuntu-rmarkdown-report

⏳ Managing EOLs w. geol: the impossible 1' Mux demo

adriens — Fri, 02 Jan 2026 21:53:35 +0000

This is a submission for the DEV's Worldwide Show and Tell Challenge Presented by Mux

❔ What I Built

We have built and maintain geol a cli to help manage, report and survey software EOLs (end-of-life), both for:

Interactive cli UX
Automates DEVOPS CI/CD pipelines

See below for previous Hacktoberfest hackathon :

Hacktoberfest: Maintainer Spotlight

adriens

Oct 29 '25

⏳geol, the cli to efficiently manage EOLs like a boss

#devchallenge #hacktoberfest #opensource #devops

3 min read

⏱️ My `1'` Pitch Video

📑 Demo

🤓 Project repository : opt-nc/geol
🍿 Youtube playlist : geol playlist
🗓️ Product Kanban

💭 The Story Behind It

Since a long time now (December 2022), I started to work on following software technical debt, security, ... at scale with:

DEVOPS in mind
Datascience & dataengineering
Datavisualizations
DEVSECOPS

Then I discovered endoflife.date website, its API.
Since that day, I contributed almost 200 PRs to add/enhance products.

Also, on a daily basis, I used hugovk/norwegianblue (aka. eol) to efficiently browse products end of lifes, for example with markdown to efficiently produce Github issues of html reports:

adriens

Dec 14 '22

⌛ Manage EoLs like a boss with endoflife.date 🛑

#showdev #security #api #opensource

3 min read

Then, years later I wondered what if I could try to push it harder and build a brand new Go based version of eol,... which I did during a 48h hours hackathon :

adriens

Jan 11 '25

⏳GitHub Copilot 1-Day Build Challenge : eol, a tiny Go client to manage eols

#devchallenge #githubchallenge #ai #go

2 min read

The result was pretty nice, I like the feeling... especially :

The "Eat your own dog food approach" (aka. Dogfeeding)
Freedom to design a brand new client, with a dedicated UX
Remove the need for Python runtime
Try some exciting new ideas
Improve productivity and talk about technical debt around me

Finally we officially created geol :

Hacktoberfest: Maintainer Spotlight

adriens

Oct 29 '25

⏳geol, the cli to efficiently manage EOLs like a boss

#devchallenge #hacktoberfest #opensource #devops

3 min read

⚙️ Technical Highlights

We wanted to learn as much and take as much profit as possible from Go ecosystem, to adopt the same stack of some tools we love - and admire - to use on a daily basis :

Language : Golang
Delivery : brew to deliver/install efficiently as fast - and yet secure - as we want
Release Management : goreleaser/goreleaser
spf13/cobra for CLI
charmbracelet/fang on top of Cobra to give glamour to our cli
charmbracelet/lipgloss for styish terminal outputs : colors, arrays, links, markdown display and redirection t files
duckdb as primary export format
osv.dev to continuously check for vulnerabilities in our base code
Code QL as part of our CI to ensure our tool does not bring security issues
Docusaurus for nice professional website (work in progress)
crush and gemini-cli to evaluate the design of our cli
pandoc for document post-processing (eg. transform output markdown into html or pdf...)
LaTeX to build the cheetsheat
The way we an put together geol and trivy (with crush or gemini-cli) at work together to build efficient and nice looking technical debt and security reports
distrobox to test install and binaries on various OS and architectures
Now, I'm starting to focus on what can be done around geol outputs to automate reporting, with a professional data-stack, like Rmarkdown or quarto to make professional looking technical debt reports
Also, I'm thinking about the most efficient way to share EOL's datas into other systems to make technical debt easier to monitor

🐦 Some Tweets showcasing what's achieved and how

`distrobox`

// Detect dark theme var iframe = document.getElementById('tweet-1987002052460106151-80'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=1987002052460106151&theme=dark" }

Putting `geol`, `trivy`, `gemini` & `latex` to work

// Detect dark theme var iframe = document.getElementById('tweet-1989079778058727889-348'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=1989079778058727889&theme=dark" }

Building cheetsheat with pure code w. `LaTeX`

// Detect dark theme var iframe = document.getElementById('tweet-1995616096351891548-64'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=1995616096351891548&theme=dark" }

Build a website with `docusaurus`

// Detect dark theme var iframe = document.getElementById('tweet-2005772435057242580-641'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=2005772435057242580&theme=dark" }

⏳geol, the cli to efficiently manage EOLs like a boss

adriens — Wed, 29 Oct 2025 20:39:59 +0000

This is a submission for the 2025 Hacktoberfest Writing Challenge: Maintainer Spotlight

⏱️ `geol` in `5'` for impatients

🤗 Building ideas & patterns before software

First of all, I would like to point what makes my core motivation & energy last, what keeps me in movement : my curiosity.
I only understood recently that this is curiosity and the satisfaction we get once got answers that (from my POV) makes things possible.

Of course we are building software or products, but first, I want to focus on creating and experimenting news ways of viewing problems that affect us and prototype new ways of fixing issues (regardless of programming languages or technologies) or rewiring known problems, then share these techniques and POVs with others to confront perspectives.

Nothing excites me more that seeing the solution come to life, talk with my teammates about how to tweak an output, what to remove until we get something really useful and pleasant.

Some call this being nerd... I don't care but what I enjoy the most is talking about software engineering, DEVOPS and architecture while enjoying a coffee, drawing something on a paper, challenging perspectives and make it happen on a schedule, at a regular pace... and gain experience.

💭 About the core idea

Back in time time in January, I submitted a first try to porting eol to a golang stack :

adriens

Jan 11 '25

⏳GitHub Copilot 1-Day Build Challenge : eol, a tiny Go client to manage eols

#devchallenge #githubchallenge #ai #go

2 min read

Also I previously talked - a lot - about these data and why it matters:

// Detect dark theme var iframe = document.getElementById('tweet-1968410713480733162-10'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=1968410713480733162&theme=dark" }

I wanted to push ~~a bit~~ much further around around stack lifecycle management and bring DEVOPS,DEVSECOPS and SECOPS to the front and imagine complementary practices to automate things around that.

After a paper and pen ideation session :

// Detect dark theme var iframe = document.getElementById('tweet-1974974775836754362-999'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=1974974775836754362&theme=dark" }

I finally opted for a full unboxing video that would focus on the MVP, from start to end.

🍿 Unboxing demo

🧑‍🤝‍🧑 Team mates

Innovation is much more interesting when it's achieved with other people, below the core team:

👨 Adrien : adriens
👨 Vinh : supervinh
👱🏻‍♀️ Michèle : mbarre

🎯 Our `MVP`

I designed a MVP milestone to be sure to be ready to deliver for D-day and at last deliver the demo:

📢 Tell the world `geol` exists #4

adriens posted on Aug 10, 2025

🎯 DoD

[ ] Write a dedicated article
[ ] Create an unboxing video
[ ] Tweet
[ ] linkedIn
[ ] Add as a contribution to https://github.com/endoflife-date/endoflife.date

View on GitHub

🤩 What we learned and techniques we developed on the journey

🧰 Stack

Charmbracelet
charmbracelet/fang "The CLI starter kit"
charmbracelet/lipgloss "Style definitions for nice terminal layouts 👄"
Cobra
gemini-cli
charmbracelet/crush "The glamourous AI coding agent for your favourite terminal 💘"
GoReleaser
go-task/task
pandoc
asciidoctor
fish-shell/fish-shell "The user-friendly command line shell."
aquasecurity/trivy "Find vulnerabilities, misconfigurations, secrets, SBOM in containers, Kubernetes, code repositories, clouds and more"

➿ `cli` design tuning with AI

I used gemini-cli and crush to continuously evaluate & improve the design of our cli : which command to add, examples to provide, better options, option names, sub-commands layouts,...

👉 Our goal was to achieve zero-prompts answers, eg. :

// Detect dark theme var iframe = document.getElementById('tweet-1970676017556865268-741'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=1970676017556865268&theme=dark" }

Ask the agent "learn to use geol thanks to man pages" (it appeared than, after benchmarking, man was the best option)
Challenge the assistant to produce new kind of documents, for example with the help of third parties cli like trivy or grype

📰 Original report productions

I challenged geol and trivy as data producers, gemini-cli as thinker to efficiently produce LaTeX some original reports pretty easily and share them:

// Detect dark theme var iframe = document.getElementById('tweet-1971740458331394521-342'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=1971740458331394521&theme=dark" }

🔭 Further ideas

Below some features to be implemented that excites us a lot:

Stack survey : 🔔 Lock, warn & report watch stack 👀
Data exports : 📤 Add export and duckdb sub-command : geol export duckdb
Documentation export : design still in progress 😅 but preparing some cool work

🙌 Last words

This project makes us learn a lot about software packaging, distribution, make choice to be sure to deliver the most ~~important~~ useful features, also carefully not implement the things that should not be... and learn learn and learn about free software.

🛡️ Data quality, SQL, duckdb and http_client on CI🦆

adriens — Mon, 08 Sep 2025 04:02:44 +0000

💭 CI, `duckdb` et and data protection

To efficiently yet effortlessly manage data quality, we created a GitHub Action to install duckdb :

adriens for opt-nc

Jul 25 '23

🦆 Effortless Data Quality w/duckdb on GitHub ♾️

#duckdb #githubactions #data #automation

1 min read

... but recently I had to face an another challenge : as part of our CI, I had the need to validate data... that were relying on web resources.

I needed to be sure that a GitHub Account was really existing (for example to avoid typos) as part of our CI.

In this very short article, I'll show how to use DuckDB with the http_client extension to verify GitHub handles stored in a table, for example to lint data as part of a CI pipeline thanks to GitHub Duckdb Action... and do the job with a very simple SQL script and CHECK constraints.

🍿 For impatients

🔖 Resources

🕹️ Do the job with `SQL`

First install extension:

INSTALL http_client FROM community;
LOAD http_client;

Then create a table with some example data:

create or replace table person_gh_member
(
  sam_accountname varchar primary key,
  gh_member varchar not null
);

-- insert rows
INSERT INTO person_gh_member (sam_accountname, gh_member)
values
('adriens', 'adriens'), 
('jdoe', 'johndoe'), 
('asmith', 'annasmithRRRRR');

Finally, run a query to check the status of each GitHub handle:

create or replace view v_person_gh_status as
    select 
    sam_accountname,
    gh_member,
    'https://github.com/' || gh_member as gh_url,
    cast(http_get(gh_url).status as integer) as http_gh_status
    from person_gh_member;
    --where gh_status <> 200;

from v_person_gh_status;

Now, to lint, we can add a check to see if any status is not 200:

from v_person_gh_status
    where http_gh_status <> 200;

create or replace table lint_gh_handle(
    gh_handle varchar primary key,
    gh_url varchar not null unique check (gh_url like 'https://github.com/%'),
    gh_status integer check (gh_status = 200)
);

insert into lint_gh_handle(
            gh_handle,
            gh_url,
            gh_status)
select
    gh_member,
    'https://github.com/' || gh_member as gh_url,
    cast(http_get(gh_url).status as integer) as gh_status
from person_gh_member;

🎬 2025 UNC MIAGE démo finale

adriens — Sun, 18 May 2025 20:57:50 +0000

❔ Une video... pourquoi pas un rapport 🤔

// Detect dark theme var iframe = document.getElementById('tweet-1926774800480870459-19'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=1926774800480870459&theme=dark" }

// Detect dark theme var iframe = document.getElementById('tweet-1915874203083149769-186'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=1915874203083149769&theme=dark" }

📜 Règles de base

Teaser < 1' : que j'utiliserai pour faire les shorts YT pour pitcher votre travail (elevator pitch)
Rappeler le cadre qui a permis ces travaux (UNC/IAE, promotion,...)
Durée totale ≤ 20'
Format : horizontal (16:9), format MP4
Texte et liens vers les ressources démontrées (datasets, notebooks, repos,...)
Résumé (5 lignes max)
Pitch résumé format twitter
Nom du fichier : 2025_unc_miage_{kaggle_handle}.mp4
Un thumbnail format YT (visuel, logo UNC/IAE, stack, avatar Kaggle,...)

🎯 Objectif

Présenter votre projet de fin d’année sous forme d’une vidéo synthétique et structurée, destinée à restituer de manière claire, convaincante et captivante plus de 3 mois de progrés et de travaux.

La vidéo devra à minima inclure :

La problématique
La collaboration mise en oeuvre
Le pipeline technique
Les résultats obtenus
Les limites et perspectives
Ce que vous auriez aimé creuser mais que vous n'avez pas pu faute de temps

💡 Idées

Se faire plaisir
Etre créatif
Profiter des possibilités ouvertes par le montage video (montage ,coupage, animations,...)
Créer un support de présentation en ligne (Canva, ppt, ...) afin de structurer le contenu, avec des visuels et des revues de code,...
Mettre en avant le caractère innovant pour la Nouvelle-Calédonie
Ce que vous avez découvert
Mettre en avant ce que vous avez aimé (côté humain, technologique,..)
Pointer là où vous vous êtes fait plaisir
Ce qui a suscité le plus de curiosité ou l'envie de continuer vers plus de recherche, d'approfondir une technologie ou un thème
Si vous pouviez passer le flambeau à d'autres étudiants l'année prochaine ce vous aimeriez voir réalisé
La mise en perspective des travaux des autres camarades de promotion
Donner du feedback sur la pédagogie ou le pipeline didactique qui a émergé
En quoi vous pensez que votre travail est innovant voire disruptif
En quoi votre travail bénéficie à la société
Contexte métier ou sociétal
Sources de données et pipeline (collecte, nettoyage, transformation)
Modèles/stack utilisés et justification
Résultats (graphiques, métriques, comparaisons)
Analyse critique, limites, éthique
Impacts concrets ou potentiel du projet
Chiffres clés ou visuel fort
Appel à l’action (“Découvrez la vidéo complète !”)
Limites & perspectives
Si vous avez dépassé les objectifs que vous vous étiez fixés
**Si/comment cette expérience va impacter **votre carrière
...

🧮 Axes d’évaluation

Capacité à captiver le spectateur
Clarté et structuration du propos
Mise en avant des problématiques d'interopérabilité d'un point de vue technique
Problématiques organisationnelles
Qualité technique de la démarche
Analyse critique des résultats et des limites
Valorisation des résultats (visualisation, impact)
Qualité visuelle et sonore de la vidéo
Capacité à vulgariser / synthétiser
Curiosité suscitée chez le spectateur pour le travail réalisé ou de nouvelles questions

🧭 Dimension	Description	Note / Pondération
1. Clarté et structuration du message	Présentation logique, fluide, introduction + développement + conclusion. Le discours est compréhensible, bien rythmé.	/4
2. Qualité de vulgarisation et de synthèse	Capacité à expliquer clairement les concepts, à vulgariser les aspects techniques, à aller à l’essentiel sans perdre de rigueur.	/4
3. Richesse et pertinence technique	Pipeline de données, stack, interopérabilité, justifications techniques, rigueur méthodologique.	/4
4. Créativité, montage et impact visuel	Originalité du montage, qualité sonore et visuelle, utilisation pertinente de visuels, démos, animations.	/4
5. Portée, analyse critique et engagement	Mise en perspective, limites, éthique, bénéfices pour la société, passion, envie de transmettre, implication personnelle.	/4

🥳 We built the cli of our dreams to send sms ❣️

adriens — Mon, 14 Apr 2025 21:48:18 +0000

🤔 About the benefits of `cli`

Since a few years now, we started to design various cli for internal batch usage, on our Java Stack on top of picocli and quarkus, delivered as images, and run on podman.

Our goal was to :

Avoid as much as possible to write unnecessary documentations : each tool should describe himself according to a common pattern
Rely on the best possible practices
And overall : deliver a consistent first-class UX to OPS

Therefore I started to read and follow the following site to follow the best possible guidelines :

cli-guidelines / cli-guidelines

A guide to help you write better command-line programs, taking traditional UNIX principles and updating them for the modern day.

Command Line Interface Guidelines

An open-source guide to help you write better command-line programs, taking traditional UNIX principles and updating them for the modern day.

This is the source code for the guide. To read it, go to clig.dev.

Join us on Discord if you want to discuss the guide, or just chat about CLI design.

Contributing

The content of the guide lives in a single Markdown file, content/_index.md The website is built using Hugo.

To run Hugo locally to see your changes, run:

$ brew install hugo
$ cd <path>/<to>/cli-guidelines/
$ hugo server

To view the site on an external mobile device, run:

hugo server --bind 0.0.0.0 --baseURL http://$(hostname -f):1313

Deployment

The site is hosted on Cloudflare Workers as a static-assets project, built and deployed from this repo:

build.sh installs a pinned Hugo and builds the site into public/.
wrangler.toml runs build.sh and uploads public/…

View on GitHub

🎯 Our `Go`(al)

On my very own side I gave a try during a hackathon to Go and goreleaser :

adriens

Jan 11 '25

⏳GitHub Copilot 1-Day Build Challenge : eol, a tiny Go client to manage eols

#devchallenge #githubchallenge #ai #go

2 min read

Then, I wanted to bring my team to the Go experience at the office as sometimes we need to deliver apps to systems on which we don't want or can't install new softwares : so delivering a static binary thanks to Go and goreleaser seemed a good option for effortless cross-compilation... and of course build the best possible UX thanks to Cobra.

So I used this opportunity to sharpen our skill with my team :

// Detect dark theme var iframe = document.getElementById('tweet-1909597554590855170-191'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=1909597554590855170&theme=dark" }

And...

Build the best possible cli of our dreams to send sms from a terminal and see what we can deliver and learn from that.

🍿 `v1.0.0` `brew` unboxing

📜 The story behind the product development

As I always say to my team & student, always "Start with why" :

// Detect dark theme var iframe = document.getElementById('tweet-1910458188945867005-547'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=1910458188945867005&theme=dark" }

After having drafted my ideas :

// Detect dark theme var iframe = document.getElementById('tweet-1908757267941666914-800'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=1908757267941666914&theme=dark" }

Here are the motivations that kept us on track :

Do things extremely :

🧩 Interoperable
🫴 Easy to integrate for anyone (even on legacy systems)
⚡ Efficient
🦥 Useful : make people save time, ie. make more with less effort
📦 Easy to deliver and install
🤩 Beautiful
📉 Fast to build : for an improved Time to market
🟢 Fast to (learn to) use
💡 Inspiring for us and others, eg. create new business opportunities ideas
😍 Desirable : people should feel the desire to use the product and feel joy while using it
🧑‍🤝‍🧑 Create new relationships with people and discover new practices

🔭 What's next

Next, we'll focus on showing and sharing the benefits :

♾️ Continuous delivery tools and automation
📼 New ways of documentation with next-gen tools
🛡️ Security (beacause it always matters)
🦸 One liners (because they're both a cool and efficient way to get things done)
🐧 Classical linux tools
📚 General IT Culture
🧑‍🎨 Creative product design and development

Hopefully you'll like it as much as we do...and will engage yourself on the same kind of way : the #learnbydoing journey 🤗

🔬Public docker images Trivy scans as duckdb datas on Kaggle

adriens — Mon, 31 Mar 2025 21:03:40 +0000

❔ About

As Trivy is becoming quite popular :

// Detect dark theme var iframe = document.getElementById('tweet-1901655379542475129-914'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=1901655379542475129&theme=dark" }

I decided to give it a first try... and see what could be achieved within a very (very) few lines of code :

With datascience
Scheduled on Kaggle
With existing datasets
Deliver nice ready-to-use data as duckdb, parquet and csv

... and last but not least produce some first dataviz and see how it goes:

But... enough talk, let's see how it looks like !

DEV Community: adriens

📊 I (finally) built the demo of my duckdb brew extension

💭 Where it comes from

🤗 What I Built

🍿 Demo

🤔 The Comeback Story

🦥 My Experience with GitHub Copilot

⚚ I reverse-engineered my mobile operator's APK — then Hermes Agent wrote the executive report

🍿 Demo

📸 Screenshots

Building a Multi-Stakeholder API Monitoring Report with Hermes Agent

🏝️ The Backstory: A Stack No One Else Has

🎯 The Problem Hermes Agent Solved

🎭 The Role-Playing Game That Became the Design Document

⚡ What Hermes Agent Built in ~1 Hour

📊 The Charts It Produced (Without Being Asked)

📋 The Executive Summary: Fits in One Slide

🧠 What I Learned

💰 The Cost

🖥️ What's Next

🧞‍♂️Transform unstructured PDFs Job Offers into a dataset w. gemma4:2b

🤔 About the power of collections and our ability to compare things

🙋 What I Built

🎯 Problems it solves

🤗 Experience it creates

🍿 Demo

💰 The benefits : then and now

⚖️ Benchmark : marker-pdf vs. marker-pdf ➕ gemma4:e2b

📊 Analytics on top of database

📜 Code

🎁 Goodies

💡 How I Used Gemma 4

🤔 What remains to do...

💸 How $0 marketing budget boosted my creativity & automation opportunities

💡 Inception back in days

⏳ About geol : the need for a logo

🎨 How low marketing skills and budget boosted my creativity

🍿 Show me what it looks like

📝 Inception and paper/pen drafts

☝️ Put more constraints to be more productive

🚀 Pure code and automation

🖼️ Benefits : the 100% LaTeX cheatsheet

🔭 Further opportunities

🦈 Analyzing Open Data shark attacks with gemini-cli and Quarto 📊

😂 Spoiler alert : what it was not

❔ What I Built with Google Gemini

🔥 Work planning (brain on fire)

🍿 Demos

Long edition with pdf commenting

Short edition (for educational purposes)

📑 Resources

🧑‍🎓 What I Learned

💡 Google Gemini Feedback

📸 Screenshots

🦆Built a duckdb community extension for brew

🤓 What I Built

💡 The idea : the duckdb extensions community

🍿 Demo

📷 Screenshots

📈 Historical brew install

🌌 Graph Dependencies

🐧 SQL reports from terminal

🤓 My Experience with GitHub Copilot CLI

👨‍🎓 Lesson learned : CI, tests and lint first

📑 Resources

⏳ Managing EOLs w. geol: the impossible 1' Mux demo

❔ What I Built

⏱️ My 1' Pitch Video

📑 Demo

💭 The Story Behind It

⚙️ Technical Highlights

🐦 Some Tweets showcasing what's achieved and how

distrobox

Putting geol, trivy, gemini & latex to work

Building cheetsheat with pure code w. LaTeX

Build a website with docusaurus

⏳geol, the cli to efficiently manage EOLs like a boss

⏱️ geol in 5' for impatients

🤗 Building ideas & patterns before software

💭 About the core idea

⚖️ Benchmark : `marker-pdf` vs. `marker-pdf` ➕ `gemma4:e2b`

⏳ About `geol` : the need for a logo

🖼️ Benefits : the `100%` LaTeX cheatsheet

💡 The idea : the `duckdb` extensions community

🐧 `SQL` reports from terminal

👨‍🎓 Lesson learned : `CI`, tests and lint first

⏱️ My `1'` Pitch Video

`distrobox`

Putting `geol`, `trivy`, `gemini` & `latex` to work

Building cheetsheat with pure code w. `LaTeX`

Build a website with `docusaurus`

⏱️ `geol` in `5'` for impatients

🎯 Our `MVP`

➿ `cli` design tuning with AI

💭 CI, `duckdb` et and data protection

🕹️ Do the job with `SQL`

🤔 About the benefits of `cli`

🎯 Our `Go`(al)

🍿 `v1.0.0` `brew` unboxing