DEV Community: David Anton Alvarez

Read the Fine Print: a contract red-flag reviewer for freelancers that never uploads your contract

David Anton Alvarez — Sun, 18 Oct 2026 15:24:45 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle

What I Built

Freelancers sign contracts written by the client's side. The traps are always the same: payment at Net 90, "unlimited revisions", IP transferred before you've been paid, no kill fee, a non-compete, unlimited liability, an auto-renewal you forget about. Most of us skim, sign, and find out later.

Read the Fine Print is a web app for that moment before you sign. Paste the contract or drop the PDF, and every clause gets a traffic light:

🔴 To change: don't sign it as it is.
🟡 To negotiate: common, but worth pushing back on.
🟢 Fine: nothing to do.

Each flagged clause comes with a plain-English explanation and a ready-to-send message asking for a fairer version. Copy email to client puts them all into one polite email.

The important part: the AI runs entirely in your browser. Client contracts are confidential and often under NDA, so the app never sends them anywhere. No server, no account, no API key.

Who would pay for it? Freelancers who sign a handful of contracts a year and can't justify a lawyer for each one. A lawyer's review costs more than many small gigs pay; a one-off purchase or a few dollars a month for a tool that catches the obvious traps is an easy sell.

Demo

Try it: virien84.github.io/read-the-fine-print (needs WebGPU: recent Chrome or Edge on a laptop/desktop). Click Try a sample contract to see it work without using a real one.

The first visit downloads the model (~3 GB) once; after that the browser caches it and it loads in about 11 seconds. Each clause takes 7–13 seconds on my NVIDIA Turing GPU, and the bundled 13-clause sample contract is reviewed in about 2.5 minutes, with live progress as each clause comes in.

On that sample, which I seeded with the classic traps, the app flags the payment terms, unlimited revisions, IP-before-payment, non-compete, liability and auto-renewal clauses. It's honest to say it isn't perfect: the termination clause only gets a yellow, and it leans strict (a fair IP clause sometimes comes back yellow instead of green). For a "read this before signing" tool, erring on the side of caution is the right failure mode.

I haven't put it in front of a paying freelancer yet, so no first-user story this week.

Code

Virien84 / read-the-fine-print

Read the Fine Print

Spot the clauses that cost freelancers money, before you sign.

Paste or drop a client contract and every clause gets a traffic light:

🔴 To change: don't sign it as it is.
🟡 To negotiate: common, but worth pushing back on.
🟢 Fine: nothing to do.

Each flagged clause comes with a plain-English explanation and a ready-to-send message asking for a fairer version, and Copy email to client puts them all in one email.

The AI runs entirely in your browser. Client contracts are confidential and often under NDA, so Read the Fine Print never sends them anywhere: no server, no account, no API key.

How it works

Piece	What it does
Gemma 4 E2B via WebLLM	Runs on your GPU through WebGPU, in a Web Worker. Downloaded once (~3 GB) and cached by the browser. Gemma 3 1B is the fallback for

…

View on GitHub

Vite + React, no backend.

How I Built It

PDF / pasted text ─► pdf.js (local) ─► split into clauses
                                            │
                                            ▼
                     Gemma 4 E2B (WebLLM, WebGPU, Web Worker)
                     JSON-schema output, one clause at a time
                                            │
                                            ▼
                       🔴🟡🟢 + explanation + pushback ─► email to client

Model: Gemma 4 E2B through WebLLM, in a Web Worker so the page stays responsive. Gemma 3 1B is the fallback for GPUs without shader-f16.
Constrained output: every answer follows a JSON schema, so parsing never fails. The trick that mattered most: putting explanation before level in the schema. When the model has to explain the clause first and only then pick a colour, its verdicts got noticeably more accurate.
A checklist of known traps: the prompt lists the classic freelance traps, so a small model compares against a list instead of relying on its own judgement. I also had to add a rule that the pushback must actually be better than the original (an early version "negotiated" a 60-day notice period up to 90).
Clause splitting: numbered clauses first, paragraph fallback, signature blocks skipped. PDFs are read locally with pdf.js.
Surviving a lost GPU: on long runs Chrome sometimes loses the WebGPU device and refuses it until reload. The app parks the review in sessionStorage, reloads, and resumes where it left off.

One lesson: more is not always better. I tried "red lines" (user-defined deal-breakers), an enum of trap types and a second yes/no verification pass. On the full sample, accuracy got worse (the preamble got flagged, revisions got missed) and it cost twice the GPU time. I went back to the single, simple prompt.

I built it with Claude Code as my coding agent.

Why Does Open Innovation Matter?

Privacy is the product. Many contracts forbid sharing them with third parties. A cloud API is a third party. An open-weight model running on your own GPU is not, so you can review an NDA-covered contract without breaking it.
The economics work. No per-token bill means I can sell it cheaply (or give it away) without losing money on every review.
No lock-in. If a better small open model ships next month, it's a one-line change.

Prior art and inspiration from the DEV community:

The 5 contract clauses that quietly wreck freelancers by agentedeia: the clause → pushback format.
The biggest mistake I made with client contracts by aldobarbosa: unlimited revisions in real life.
OCR + LLM Pipeline for Contract Review by methezain.
I built a local LLM that runs entirely in your browser by zaydmulani09.
Building a Context-Aware Legal Contract Analyzer with Gemma and Gemini by __13_77. My angle: fully local, and focused on freelancers.

Not legal advice. For anything high-stakes, ask a lawyer.

Prize Categories

Best Use of Gemma: Gemma 4 E2B runs fully in the browser through WebGPU and does the whole review, with JSON-schema constrained output per clause.

NaturDex: a Pokédex for real nature that runs offline on a recycled phone

David Anton Alvarez — Sun, 11 Oct 2026 01:05:05 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass

What I Built

We live surrounded by plants and animals we never look at. The robin on the fence, the heather on the hillside, the butterfly crossing the path. Kids walk past all of them with their eyes on a screen.

So I flipped it around and used the screen to send them outside. NaturDex is a Pokédex for real nature. You photograph a plant or an animal, and an AI model running on the phone tells you what it is. If it's new, it goes into your NaturDex with its number, its photo and the date you found it. The goal is to fill it, and the only way to do that is to go out and look.

The scenario I designed it for is a school field trip to the forest. Schools could hand out recycled phones (old phones everyone has in a drawer) with NaturDex installed, and the activity is to fill the NaturDex. That scenario set three hard rules:

No signal. Forests don't have coverage, so everything has to work offline.
Kids' privacy. No accounts, no uploads. Photos and GPS positions of children never leave the phone.
Old hardware, zero cost. It has to run on a phone that's a few years old, with no app store and no API bill.

And, since it's for children outdoors, a fourth one: safety first. NaturDex warns before it entertains.

How a trip works:

Before leaving, on wifi: tap Prepare trip. It downloads the model (87 MB) and the list of 1,944 species that live in the region (Asturias, in northern Spain).
In the forest: tap the big camera button. In a few seconds NaturDex answers in one of three ways: "It's a European Robin!", "Which one is it?" (and the kid compares the photo with three options), or "Hmm, I'm not sure…" with a tip to take a better photo.
Every species found gets its NaturDex number (No. 0046 for the robin) and a coloured badge for its kind: bird, plant, butterfly, fungus…

Why only Asturias? Because it's where I live, and where I could actually use it. The AI model knows living things from all over the world, but the species list and, above all, the safety rules have to be right for the place where the kids are. Which vipers live here, which plants are poisonous, which species are protected by law: I can only vouch for that in a region I know. So I chose to do one region well instead of the whole world badly. The last section explains how other regions could join.

If you don't live in Asturias, you can still try it: on a computer, the camera button opens a file picker, so you can give it any photo of a common European species, like a robin, a fly agaric or a peacock butterfly.

Demo

Try it: virien84.github.io/naturdex. Open it on a phone, tap Prepare trip on wifi, then go outside. You can also install it from the browser menu (Install app / Add to Home screen).

The design is a handheld field device: a red shell, a big blue lens that glows while the AI is thinking, a leaf-green LCD screen and chunky buttons that press down.

Ready to go	A confident answer	Added to the NaturDex

That robin is a small bird far away, between branches, and NaturDex is still 100% sure. The shield means it's a protected species: look, but don't disturb it.

Safety comes first. Dangerous species get a black-and-yellow hazard warning above everything else. On the right, the model isn't sure which snake it is, but one of the candidates is a snake, and that's enough to warn.

Deadly mushroom	"Could it be…?"

When the model hesitates, the kid decides. It's a learning moment: compare your photo with the options. And when NaturDex really doesn't know, it says so instead of guessing.

Which one is it?	Not sure	My NaturDex

At night the screen switches to a backlit LCD:

The photos in these screenshots are openly licensed iNaturalist observations from Spain (CC BY / CC0). Credits are in screenshots/CREDITS.md.

I took it outside

I installed it on my phone, a mid-range Android (vivo V50 Lite), and went out to photograph trees and plants. All three answers showed up: direct identifications, "Which one is it?" with the right species among the options, and "not sure". It worked with no errors.

The "not sure" one taught me something. I had photographed a whole tree standing among other trees. The model looks at a square crop of 224×224 pixels, and in that square there were several crowns and a lot of sky, so it couldn't tell which tree I meant. It was right not to guess. So now the camera says "For trees, photograph one leaf, flower or the bark", and the "not sure" screen repeats it.

The second lesson was about patience. After taking a photo, the app looked frozen. The message "Looking closely…" existed, but the screen never got the chance to show it (more on that below). Kids would tap again, or give up. Now the lens glows and a spinner turns from the very first instant.

Code

Virien84 / naturdex

NaturDex

A nature "Pokémon Snap" for school field trips: photograph plants and animals and fill your NaturDex, fully offline on recycled phones.

Try it: https://virien84.github.io/naturdex/ (open it on a phone, tap Prepare trip on wifi, then go outside).

app/ — the PWA (React + Vite + vite-plugin-pwa + onnxruntime-web).
embeddings/ — Python scripts (uv) that prepare the model and the species list on a laptop.
SAFETY.md — how NaturDex keeps kids and protected species safe.

How it is hosted

What	Where
The app	GitHub Pages, rebuilt on every push to `main` by `.github/workflows/deploy.yml`
The BioCLIP model (87 MB, too big for GitHub)	Hugging Face: Virien/naturdex-bioclip-onnx
Species list + embeddings (~5 MB)	In this repo, `app/public/species/`

Everything runs on the phone: photos and locations never leave it.

Credits

BioCLIP by the Imageomics Institute (MIT).
Species occurrences from GBIF; test photos from iNaturalist.

View on GitHub

app/: the PWA (React + Vite + vite-plugin-pwa + onnxruntime-web).
embeddings/: Python scripts that prepare the model and the species list on a laptop.
SAFETY.md: every safety rule and where the data behind it comes from.

MIT licensed. The model is on Hugging Face: Virien/naturdex-bioclip-onnx.

How I Built It

The trick that makes it fit on a phone is splitting the AI work in two: the heavy, one-off part runs once, on a laptop, and the phone only does what it has to do for each photo.

LAPTOP, once                                 PHONE, every photo (offline)

GBIF ─► 1,944 species of Asturias            Photo ─► BioCLIP image encoder
          │                                          (ONNX int8, 87 MB, WASM)
          ▼                                              │
BioCLIP text encoder ─► 1 "fingerprint"                 ▼
per species (512 numbers)  ──────────────────►  compare with the 1,944
                                                fingerprints ─► top 5 + %

1. The model: BioCLIP. BioCLIP is an open CLIP model (MIT) trained by the Imageomics Institute on the Tree of Life: millions of images of organisms with their full taxonomy. Like any CLIP model, it puts images and texts in the same space, so it can do zero-shot classification: no training on my side, just compare the photo with a text description of each species.

2. The species list, on the laptop. A script downloads from GBIF every species with at least 20 human observations in Asturias since 2000 (1,933), and a safety script adds 11 dangerous species that are too rare to make the cut, because the app can only warn about species it knows. Then BioCLIP's text encoder turns each species into a "fingerprint": its full taxonomy and common name, averaged over 15 photo prompts ("a photo of a…", "a blurry photo of a…"). That's a 4 MB file.

3. The model, on the phone. I exported only BioCLIP's image encoder to ONNX and quantized its weights to int8: from 345 MB to 87 MB. It's too big for GitHub Pages, so it lives on Hugging Face, which serves it with CORS. The phone runs it with onnxruntime-web (WebAssembly), compares the photo's embedding with the 1,944 fingerprints, and picks a verdict:

const top = candidates[0].probability
const verdict = top >= CONFIDENT ? 'confident' : top >= UNSURE ? 'maybe' : 'unsure'
return { candidates, verdict, warnings: precautionWarnings(candidates) }

Those thresholds (70% and 40%) aren't guesses. I evaluated the int8 model, running exactly as the phone does, on 244 research-grade iNaturalist photos from Spain (82 species, including every dangerous one):

top-1	top-3	top-5
66%	81%	86%

In the evaluation (an earlier run, with 232 of those photos), above 70% the top answer was right 84% of the time, so NaturDex says "It's a…!". Between 40% and 70% it was right only ~45%, but the right species was usually in the top 3, so NaturDex shows three and lets the kid choose. Below 40% it was right 7% of the time, so it says "not sure".

4. Safety rules (SAFETY.md). This is the part that matters most, because a wrong answer here isn't a bug, it's a child touching something.

Look, don't touch. Every species can carry a warning: danger (vipers, the Asian hornet, foxglove, death cap…) or caution (wasps, nettles…). They come from a hand-curated list with a source for each, plus group rules: never eat wild mushrooms, don't touch amphibians' skin.
Precaution rule: if any of the top 5 candidates with at least 5% probability is dangerous, the warning shows, even when it's not the top answer. That's the "Could it be a Viperine Snake?" screenshot. On the 124 test photos of hazardous species, 118 got a warning.
Every snake is "danger". In the test, the model identified a venomous Baskian viper as a harmless viperine snake with 96% confidence. Kids can't tell them apart either, so NaturDex never relies on that difference.
NaturDex never says that anything is edible.
Hide where threatened species live. The 283 species protected in Spain or Asturias show a "look, but don't disturb it" note, and for threatened ones the saved GPS position is rounded to about 10 km, like iNaturalist does, so a shared album can't lead anyone to a nest.

5. Offline for real. It's a PWA: a service worker caches the app shell, the fonts and onnxruntime's WebAssembly; the model and species list go to Cache Storage when you tap Prepare trip; the album lives in IndexedDB. Once prepared, it works with no connection: I tested it by preparing the trip, cutting the network, reloading the page and identifying a butterfly. If you're building something similar, this post on offline PWA pitfalls by ohugonnot covers the Vite PWA and Workbox traps well.

Three bugs worth sharing

The app looked frozen, because the browser never got to paint. Decoding a full-size phone photo, checking whether it's blurry and running the model all happen on the page's single thread. React had already switched to the "thinking" screen, but the browser had no free moment to draw it. The fix is to wait for a paint before the heavy work:

// Waits until the browser has drawn the screen. Decoding the photo and running
// the model block the page, so without this the "thinking" screen never shows.
function nextPaint() {
  return new Promise((resolve) => requestAnimationFrame(() => setTimeout(resolve, 0)))
}

The spinner and the glowing lens are CSS transform/opacity animations, which the GPU keeps running even while the model has the page busy. Measured inside the page with the CPU throttled 4×: the spinner is painted 0.05 s after the photo arrives, and the result comes ~13 s later.

Firefox got stuck at 99%. After downloading, the app asks the browser not to delete its data (navigator.storage.persist()). Chrome, Brave and Safari answer at once. Firefox asks the user first, and the app waited for that answer. If the kid didn't notice the prompt, the download looked stuck forever. Now it waits at most 3 seconds and moves on. I found it by running the full flow (download + identify) in Chromium, Brave, Firefox and WebKit with Playwright, and all four pass now.

"It's an European Robin!" In English it's a European. Small, but the app talks to kids.

I built it with Claude Code as my coding agent: it did the heavy lifting on the ONNX export, the evaluation script, the PWA plumbing and the design, while I made the decisions and tested it outside.

Why Does Open Innovation Matter?

For this project, open isn't a nice-to-have. A closed API would have broken every one of the rules:

It works in the forest. A cloud model needs coverage. BioCLIP runs on the phone, in the browser, in airplane mode. That's only possible because the weights are open: I could export them to ONNX, quantize them to int8 and host the result myself.
Children's photos and locations stay on the phone. With a cloud API, every photo a child takes (and where they took it) would go to a server the school doesn't control. Here there are no accounts, no uploads and no analytics. Nothing to consent to, nothing to leak.
It costs nothing per photo. A class of 25 kids taking photos all morning would be thousands of API calls. With an open model running locally, the cost is zero, forever, and the school doesn't need a credit card or an API key.
It runs on old phones. It's just a web page: no app store, no installation process, no minimum Android version beyond a modern browser. Recycled phones are enough.
A specialist beats a generalist here. BioCLIP was trained by researchers specifically on the Tree of Life, with taxonomy. It knows thousands of European species, from beetles to orchids. And being open, I could measure exactly how good it is on photos from my region, and set the thresholds from that data instead of trusting a black box.
It's adaptable. The model doesn't change from one place to another; only the species list and the safety rules do. Another region is a matter of re-running the Python scripts with another GBIF area. No one has to ask for permission.

The data is open too: species from GBIF, test photos from iNaturalist, protected species from public Spanish catalogues. Open models plus open data made a tool that a school can own.

What's next: regions made by the people who live there

Asturias is a starting point, not the limit. The final idea, and the reason it had to be open source, is that anyone could add their own region:

A region pack is small: the species list and its fingerprints (about 5 MB, generated with the scripts in the repo from GBIF data), plus the region's safety files. The model stays the same for everyone.
The safety files are the valuable part, and the part I can't write for other places: which species are dangerous there, which ones are protected, and where to hide locations. Today safety.py is built around Spanish and Asturian catalogues. The plan is to turn it into one folder per region (dangerous.csv, sensitive.csv, the local protected-species list), so the people who know each place are the ones who feed it: a biology teacher, a park ranger, a local naturalists' group.
New regions would arrive as pull requests to a shared repository, where others can review the sources before a pack goes live, and NaturDex would list them in Prepare trip so a school picks its own before the trip.

That's the kind of collaboration a closed app can't have: the AI is open, the data is open, and the knowledge that keeps kids safe comes from the community itself.

My Agent Session

This is the session from the last day: testing it outside on my phone, fixing the "frozen" screen, the Pokédex-style redesign, the install icons, the cross-browser test that caught the Firefox bug, and preparing this post (including swapping out a photo of a run-over fox for a living one). It's in Spanish, my native language.

NaturDex: field-testing, a Pokédex-style redesign and cross-browser fixes

You

He hecho una pequeña prueba solo con arboles y plantas. He tenido los tres supuestos: no detección, detección con opción de que el usuario elija y la detección directa.

Agent

▸ Bash Read project memory and find classification thresholds

Input

$ Read project memory and find classification thresholds

Output

▸ Bash Read the classifier decision logic

Input

$ Read the classifier decision logic

Output

¡Buena señal! Los tres resultados que has visto son justo los que diseñamos. Salen de classifier.js, que mira lo seguro que está el modelo de su primera opción:

Lo que viste	Confianza de la 1ª opción	Qué tal acertó en las pruebas del PC (244 fotos)
Detección directa ("It's a …!")	≥ 70 %	Acierta ~84 % de las veces
Elegir entre opciones ("Which one is it?")	entre 40 % y 70 %	La 1ª acierta ~45 %, pero la buena suele estar entre las 3 primeras
No detección ("Not sure, take another photo")	< 40 %	Solo acertaría ~7 %, así que es mejor no …

You

1-Si. 2-No estoy seguro pero creo que si. 3-Igusl que la 2. 4-un árbol entero cerca de otros. 5-no

Agent

▸ Bash Find camera screen messages and hints

Input

$ Find camera screen messages and hints

Output

▸ Bash Find where the not-sure message is rendered

Input

$ Find where the not-sure message is rendered

Output

▸ Edit app/src/components/SpeciesCard.jsx

Input

app/src/components/SpeciesCard.jsx

Output

▸ Edit app/src/screens/CameraScreen.jsx

Input

app/src/screens/CameraScreen.jsx

Output

▸ Bash Build the app and run the linter

Input

$ Build the app and run the linter

Output

¡Perfecto! Con esto la prueba en el vivo sale bien: funciona en un móvil real, sin errores, y aparecen los tres casos.

Sobre el caso que no detectó (un árbol entero junto a otros): era lo esperable. En el recorte de 224×224 se mezclan las copas de varios árboles con el cielo, y el modelo no sabe en cuál fijarse. La app se comportó bien: prefirió decir "no estoy seguro" antes que inventarse una especie.

He hecho un cambio pequeño para guiar a los niños (todo en inglés, como el resto de la app):

Cámara (app/src/screens/CameraScreen.jsx): el consejo que sale debajo del botón ah …

You

Antes de subir y desplegar hay algo que no me gusta: después de sacar la foto, no da muestras de hacer nada y pienso que se ha colgado. Reparalo poniendo un indicador de espera

Agent

Ya veo la causa. Te lo explico y lo arreglo.

▸ Bash Check existing styles and onnxruntime settings

Input

$ Check existing styles and onnxruntime settings

Output

Ahora hago los cambios:

▸ Edit app/src/screens/CameraScreen.jsx

Input

app/src/screens/CameraScreen.jsx

Output

▸ Edit app/src/index.css

Input

app/src/index.css

Output

▸ Bash Check preview size, build and lint

Input

$ Check preview size, build and lint

Output

Compila bien. Ahora lo pruebo en un navegador automático: le paso una foto y compruebo que la ruedecita aparece antes de que salga el resultado.

▸ Bash Start the dev server for the test

Input

$ Start the dev server for the test

Output

▸ Bash Run the headless spinner test

Input

$ Run the headless spinner test

Output

Me he equivocado: he dejado en el comando una llamada suelta a python3, que se queda esperando para siempre (la memoria ya avisaba de esto). La detengo y lo repito bien.

▸ Bash Run the in-page timing test

Input

$ Run the in-page timing test

Output

photo            +0.00s
spinner painted  +0.05s
result           +12.89s

Ya está arreglado: la ruedecita de espera aparece en cuanto haces la foto. En una prueba con el procesador ralentizado 4 veces para imitar un móvil modesto, se pintó a los 0,05 s de llegar la foto, y el resultado salió a los ~13 s.

Por qué parecía colgada. JavaScript en una página web hace todo con un solo "trabajador". Después de la foto le tocaban tres tareas pesadas seguidas:
1. Abrir la foto.
2. Comprobar si está borrosa u oscura.
3. Pasar el modelo de IA.

Para redibujar la pantalla, ese mismo trabajador tiene que estar libre. Como no paraba ni un momento, el mensaje "Looking clo …

You

Haz el commit y el push

Agent

▸ Bash Commit the changes and push to GitHub

Input

$ Commit the changes and push to GitHub

Output

▸ Bash Wait for the GitHub Pages deploy to finish

Input

$ Wait for the GitHub Pages deploy to finish

Output

Ya está subido y publicado. La web https://virien84.github.io/naturdex/ ya tiene la ruedecita de espera y los consejos para fotografiar árboles.

Commit: d8c635b, "Show a spinner while identifying, and tips for photographing trees".
Despliegue: GitHub Actions lo terminó sin errores y he comprobado que la web carga bien. …

You

Mientras tengo una pregunta de Claude code: la gente recomienda instalar los plugins omni-route, frontend-design, headroom, mem-palace y playwright. Quiero instalarlos, que se puedan usar en todos mis proyectos y que me digas como se usan

Agent

Antes de instalar nada miro qué es cada uno y de dónde sale. Algunos son oficiales de Anthropic y otros son de terceros, y un plugin puede ejecutar código en tu PC.

▸ Bash List plugin subcommands and search the official marketplace

Input

$ List plugin subcommands and search the official marketplace

Output

▸ WebSearch OmniRoute Claude Code plugin github

Input

OmniRoute Claude Code plugin github

Output

▸ WebSearch headroom Claude Code plugin context compression github

Input

headroom Claude Code plugin context compression github

Output

▸ WebSearch MemPalace Claude Code plugin install github

Input

MemPalace Claude Code plugin install github

Output

▸ WebFetch https://github.com/MemPalace/mempalace

Input

https://github.com/MemPalace/mempalace

Output

▸ WebFetch https://github.com/chopratejas/headroom

Input

https://github.com/chopratejas/headroom

Output

▸ WebFetch https://github.com/diegosouzapw/OmniRoute

Input

https://github.com/diegosouzapw/OmniRoute

Output

▸ Bash Inspect the plugins' files before installing

Input

$ Inspect the plugins' files before installing

Output

▸ Bash Install the two official plugins for all projects

Input

$ Install the two official plugins for all projects

Output

Ya tienes instalados para todos tus proyectos los dos plugins oficiales, frontend-design y playwright. Los otros tres no los he instalado todavía: al investigarlos he visto cosas que conviene que sepas antes de decidir. …

You

Vamos a trabajar en el frontend, ya que es de noche y no voy a salir a sacar fotos, pero se puede trabajar en eso. Usa frontend-design para que la aplicación tenga un diseño similar a una pokedex, centrada en el público juvenil

Agent

Esta sesión empezó antes de instalar el plugin, así que leo directamente las instrucciones de la skill frontend-design y las aplico. Luego reviso cómo está hecha la app ahora.

▸ Bash Read the frontend-design skill instructions

Input

$ Read the frontend-design skill instructions

Output

▸ Bash List species kinds, keys and dependencies

Input

$ List species kinds, keys and dependencies

Output

▸ Bash Install the two font packages

Input

$ Install the two font packages

Output

Ahora el código. Primero, una función para numerar las especies como en una Pokédex:

▸ Edit app/src/ml/classifier.js

Input

app/src/ml/classifier.js

Output

▸ Write app/src/components/KindChip.jsx

Input

app/src/components/KindChip.jsx

Output

Ahora la carcasa del aparato (App.jsx), la ficha de resultado, que pasa a parecer una entrada de la Pokédex, y la pantalla del álbum (My NaturDex).

▸ Edit app/src/screens/DexScreen.jsx

Input

app/src/screens/DexScreen.jsx

Output

Ahora la hoja de estilos completa: colores, carcasa, lente, pantalla LCD, botones físicos y etiquetas de tipo.

▸ Bash Update theme colors, cache fonts, new icon, build and lint

Input

$ Update theme colors, cache fonts, new icon, build and lint

Output

Compila bien. Ahora hago capturas en tamaño móvil para revisar el diseño con mis propios ojos, como pide la skill:

▸ Bash Take light-mode screenshots of every screen

Input

$ Take light-mode screenshots of every screen

Output

▸ Bash Shrink the photo preview and remove the meter

Input

$ Shrink the photo preview and remove the meter

Output

▸ Bash Take dark-mode screenshots

Input

$ Take dark-mode screenshots

Output

En modo oscuro, el fondo de la foto sale verde claro y deslumbra. Lo cambio a oscuro fijo, cierro el servidor que se había quedado abierto y compilo:

▸ Bash Fix preview background, stop old server, build and lint

Input

$ Fix preview background, stop old server, build and lint

Output

NaturDex ya tiene un diseño tipo Pokédex: la app parece un aparato de explorador. …

Qué ha cambiado
- Carcasa roja con la gran lente azul y tres LEDs. Mientras la IA piensa, la lente brilla y …

You

Si, haz commit y push

Agent

▸ Bash Commit the redesign and push to GitHub

Input

$ Commit the redesign and push to GitHub

Output

▸ Bash Wait for the deploy and check the live site

Input

$ Wait for the deploy and check the live site

Output

El nuevo diseño ya está en la web: https://virien84.github.io/naturdex/. …

Recuerda el plazo: la entrega es mañana lunes 12 a las 08:59. Lo que queda, por orden de importancia:
1. Post de DEV.
2. Iconos para instalar la app.
3. Logros (gamificación).

You

Vamos con los iconos

Agent

Creo los iconos a partir de la lente del favicon. Hacen falta tres tamaños:
- 192 y 512 px: los que Android exige para poder instalar la app.
- Versión "maskable" de 512 px: Android recorta el icono en círculo, gota u otra forma según el móvil. Esta versión va a sangre, con el fondo rojo hasta los bordes, para que el recorte nunca deje esquinas raras.
- 180 px: para iPhone.

▸ Bash Render the PNG icons into the app

Input

$ Render the PNG icons into the app

Output

▸ Edit app/vite.config.js

Input

app/vite.config.js

Output

▸ Bash Build and check icons in HTML, manifest and offline cache

Input

$ Build and check icons in HTML, manifest and offline cache

Output

▸ Bash Ask Chrome whether the app is installable

Input

$ Ask Chrome whether the app is installable

Output

installability errors: []
manifest errors: []

Los iconos están hechos y Chrome ya considera NaturDex instalable: su comprobación interna no devuelve ningún error. …

You

Antes del commit y push: yo uso Brave, y habrá gente que use otros navegadores que no sean Chrome. ¿Funcionará igual?

Agent

Buena pregunta. En vez de responder de memoria, lo pruebo: Playwright puede usar los motores de Firefox y de Safari (WebKit), además de Chrome, y miro si tienes Brave instalado en el PC para probarlo también.

▸ Bash Download the Firefox and WebKit test engines

Input

$ Download the Firefox and WebKit test engines

Output

▸ Bash Run the full flow in Chrome, Brave, Firefox and WebKit

Input

$ Run the full flow in Chrome, Brave, Firefox and WebKit

Output

Brave, Chrome y WebKit (Safari) funcionan. Firefox se queda atascado en la descarga, así que miro qué pasa:

▸ Bash Debug the download in Firefox

Input

$ Debug the download in Firefox

Output

Downloading Species fingerprints…  99%

▸ Edit app/src/data/tripData.js

Input

app/src/data/tripData.js

Output

De paso, en las pruebas salió "It's an European Robin!". En inglés es "a European", porque "European" suena con /j/. Lo arreglo también:

▸ Edit app/src/ml/classifier.js

Input

app/src/ml/classifier.js

Output

▸ Bash Rebuild and rerun all four browsers

Input

$ Rebuild and rerun all four browsers

Output

Funciona igual en Brave, Firefox y Safari, después de arreglar un fallo que dejaba la app colgada en Firefox.

Lo he probado de verdad, no de memoria. En cada navegador hice el recorrido completo con la versión de producción: descargar el modelo desde Hugging Face, hacer una foto de un petirrojo e identificarla. …

You

Si, haz commit y push

Agent

▸ Bash Commit icons and fixes, then push

Input

$ Commit icons and fixes, then push

Output

Los iconos y los dos arreglos ya están publicados en https://virien84.github.io/naturdex/. …

You

Empecemos

Agent

Primero cargo las instrucciones para publicar en DEV y consulto las bases del reto: plantilla obligatoria, etiquetas y criterios de valoración.

▸ WebFetch https://dev.to/challenges/hacktoberfest-week1-2026-10-05

Input

https://dev.to/challenges/hacktoberfest-week1-2026-10-05

Output

▸ Bash Read the project README and safety rules

Input

$ Read the project README and safety rules

Output

Perfecto. Hago las capturas desde la web publicada, en tamaño móvil y en JPEG para que pesen poco:

▸ Bash Take post screenshots from the live site

Input

$ Take post screenshots from the live site

Output

El script tenía un fallo: después de guardar una especie, pulsaba el botón de cámara de abajo, pero si ya estás en la cámara ese botón no hace nada. Eso también le pasaría a un niño. Arreglo la app para que pulsar un botón siempre lleve al principio de esa pantalla.

▸ Bash Make pad buttons restart their screen, then build

Input

$ Make pad buttons restart their screen, then build

Output

▸ Bash Retry the post screenshots

Input

$ Retry the post screenshots

Output

Antes de seguir, dos problemas con las fotos:
1. La del zorro es un zorro atropellado: no es lo que queremos enseñar a niños ni a los jueces.
2. Las fotos de prueba vienen de iNaturalist, y cada una tiene su propia licencia y su autor. Si salen en el post y en el repo hay que citarlos.

▸ Bash Look up photo licenses on iNaturalist

Input

$ Look up photo licenses on iNaturalist

Output

▸ Bash Download openly licensed photos with credits

Input

$ Download openly licensed photos with credits

Output

Fotos con licencia libre descargadas: 32, cada una con su autor. Ahora paso todas por la app para elegir las que dan cada caso (seguro, aviso de peligro, "Which one is it?"):

▸ Bash Classify every open photo in the app

Input

$ Classify every open photo in the app

Output

You

Por favor, que no aparezca el zorro atropellado

Agent

Entendido: el zorro atropellado no saldrá. Era una foto del primer lote y ya la he descartado. Antes de elegir sustituto, miro las fotos de zorro nuevas para asegurarme de que son de un zorro vivo:

▸ Bash Retake post screenshots with open photos

Input

$ Retake post screenshots with open photos

Output

En la culebra parece verse la tripa hacia arriba, lo que podría indicar que está muerta. Con la petición que me has hecho, prefiero no arriesgar: miro las otras fotos de culebra.

Las dos están vivas. Uso la del pedregal, de Martiño Cabana Otero (CC BY), que también da "Could it be a Viperine Snake?":

▸ Bash Retake screenshots with the live snake photo

Input

$ Retake screenshots with the live snake photo

Output

Las capturas están bien. Hasta el petirrojo, que sale pequeño y entre ramas, lo reconoce al 100 %. Ahora creo el archivo de créditos de las fotos para el repositorio:

▸ Bash Commit the fix and screenshots, then push

Input

$ Commit the fix and screenshots, then push

Output

Ahora guardo esta sesión en DEV para incrustarla en el post, como hiciste en el anterior.

Prize Categories

None. NaturDex doesn't use any partner technology, so I'm entering only the overall prize.

Should I Buy: a traffic light for my friend who never makes it to the end of the month

David Anton Alvarez — Sun, 04 Oct 2026 21:26:36 +0000

This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend

What I Built

I have a friend who never makes it to the end of the month. He isn't reckless. He just doesn't know, at the moment he's looking at something in an online shop, whether he can actually afford it today. The salary arrived on the 1st, rent and bills are gone, and somewhere around the 20th the money runs out.

Banking apps tell you what you've already spent. He needed something that answers the question before he clicks "Add to cart".

So I built Should I Buy, a browser extension for Chrome and Edge. When you open a product page in any online shop, a small card appears in the corner with a traffic light:

🟢 You can afford it. It fits your spending pace for the month, and it shows how much you'd have left per day until payday.
🟡 You can, but it's tight. You have the money, but you're ahead of your pace. It tells you how many days to wait until it fits.
🔴 Not now. It tells you how many months it would take to buy it without touching your savings.

On top of the numbers, a local model writes one short, honest sentence the way a friend would say it.

You configure it once with four numbers: monthly net income, fixed expenses, how much you want to save, and your payday. Everything else happens on its own while you browse.

Demo

Settings. Four numbers, and you immediately see how much free money you really have each month.

Thinking… As soon as a page looks like a shop, the card appears while Gemma reads the page.

🟢 Green. A pack of spoons for €3.99: easy yes. Note that Gemma ignored the €13.03 shipping line and the €0.04/unit price and picked the actual price.

🟡 Yellow. Headphones for €179.99 early in the month. Affordable, but at €7.86/day for the next 28 days it's tight, and waiting 10 days makes it fit the pace. (This is my demo shop, which deliberately has a crossed-out price, a shipping cost, an instalment price and related products to try to confuse the model.)

🔴 Red. A €1,207.72 laptop with €400 of free money a month: 3 months putting aside all your free money, or 5 putting aside half. No lecture, just the math and a nudge.

The I bought it button logs the purchase, so the next verdicts take it into account.

What my friend said

I handed it over and he used it while browsing. Two things stood out to him:

It doesn't feel slow. While he's reading the product's features, the model is already thinking, so by the time he's ready to decide, the verdict is there. The wait happens while he's busy reading anyway.
He really appreciated the realistic comment and the estimate. A plain "no" would be useless; "3 months without touching your savings" is a plan.

Code

Virien84 / Should-I-Buy

Should I Buy 💸

English · Español

English

Can I afford it? A browser extension that shows you a traffic light whenever you're looking at a product in any online shop:

🟢 You can afford it: it fits your spending pace for the month.
🟡 You can, but it's tight: it tells you how much you'd have left per day until payday, and how many days to wait for it to fit your pace.
🔴 Not now: it tells you how many months it would take to buy it without touching your savings.

I built it for a friend who never makes it to the end of the month.

Screenshots

🟢 You can afford it	🟡 You can, but it's tight	🔴 Not now

Thinking…	Settings

Why open, local AI

Should I Buy sees every shop you visit and knows your income, your expenses and your savings…

View on GitHub

Plain JavaScript, Manifest V3, no build step. MIT licensed.

How I Built It

The whole thing is three pieces: a content script that decides whether a page is worth looking at, a background worker that talks to the model, and a small budget module.

Shop page ──► content.js ──► background.js ──► Gemma 3 (Ollama, localhost)
                                   │
                                   └──► budget.js (exact math) ──► 🟢🟡🔴

1. Gemma reads the shop, so I don't have to write a scraper per site.
Every shop marks up prices differently: crossed-out prices, unit prices, shipping, instalments, "customers also bought"… Instead of writing selectors for each site, the extension sends the visible text of the page to Gemma 3 4B running locally with Ollama, and asks for structured JSON using Ollama's format parameter:

const EXTRACT_SCHEMA = {
  type: "object",
  properties: {
    is_product_page: { type: "boolean" },
    product_name: { type: "string" },
    price: { type: "number" },
    currency: { type: "string" },
  },
  required: ["is_product_page", "product_name", "price", "currency"],
};

The prompt asks it to decide whether this is one product page (not a listing, search results or cart) and to return the price the buyer would actually pay. In my tests it correctly rejected a search results page, picked €179.99 over the crossed-out €249.99, and picked $749 over a "List Price" of $899.99 on an English page. It works the same on Spanish and English Amazon pages. If a shop publishes clean schema.org/Product data, the extension uses it directly and skips the model call.

To avoid sending every page to the model, the content script only wakes it up when the page looks like a shop: product structured data, a product og:type, or a price next to an "add to cart"-style button (in several languages).

2. The math is NOT done by the AI.
This was a deliberate choice. A language model is great at reading a messy page and bad at being trusted with your money, so the verdict is plain, tested JavaScript:

Free money = income − fixed expenses − savings.
Spending pace: free money is "earned" day by day from payday. On day 3 you haven't earned what you have on day 28, so a big purchase early in the month turns yellow even if the money is technically in the account.
Red verdicts compute how many months of free money you'd need, so the answer is always "without touching your savings".

const onPace = (free * cycle.elapsed) / cycle.days - spent;
if (price <= onPace) verdict = "green";
else if (price <= free - spent) verdict = "yellow";
else verdict = "red";

3. Gemma writes the friendly sentence, with guardrails.
A second, non-blocking call gives Gemma the verdict and the computed facts, and asks for one casual sentence of 25 words max, with the instruction to use only the numbers it's given. The traffic light appears first and the sentence fills in a few seconds later.

Making it fast enough on a normal laptop. On my laptop (16 GB RAM, CPU inference), a verdict takes about 9 seconds and the sentence about 6. The very first call after starting the browser took around 90 seconds, because the model has to load into memory. So the extension pre-loads the model when the browser starts, and keeps it loaded for 30 minutes (keep_alive). That's why my friend found it quick: by the time he's done reading the product, it's ready.

One gotcha worth sharing: Ollama rejects requests from browser extensions by default (HTTP 403). You have to allow them with OLLAMA_ORIGINS="chrome-extension://*" and restart Ollama. That's in the README for Windows, macOS and Linux.

I built it in an evening, pairing with Claude Code as my coding agent.

Why Does Open Innovation Matter?

Think about what this extension sees: every shop you visit, plus your income, your fixed expenses and your savings. That is about as personal as data gets. I would not ask my friend to send that to a cloud API, and he wouldn't agree to it anyway.

With an open-weight model running locally:

Nothing leaves his computer. No page text, no prices, no salary. The only network call goes to localhost:11434. His settings live in the extension's local storage.
It costs nothing to run. Someone who struggles to make it to the end of the month shouldn't need a paid API key to be told whether he can afford a pair of headphones. And because it calls the model on every product page, a per-request API bill would add up fast.
No account, no key, no sign-up. Install Ollama, pull the model, load the extension.
The model is swappable. If his laptop struggles, he can switch to gemma3:1b in the settings. If a better open model comes out, it's a one-line change.

A closed API would have made this either a privacy problem or a subscription. With an open model it's just a tool he owns.

Running locally isn't automatically private, though (this post is a good reminder). Ollama only listens on localhost, and the extension only has host permission for that address. It's in the same spirit as PocketCFO, which also keeps personal finance data on-device with Gemma.

My Agent Session

Here is the full session with my coding agent, from "what hackathons are open this month?" to the finished extension. It's mostly in Spanish, my native language. You can see the idea taking shape, the disk filling up halfway through the model download (and moving Ollama's models to another drive), the 403 from Ollama when an extension calls it, testing Gemma against tricky price layouts, and the switch to English for the judges.

Building Should I Buy: a local-AI budget traffic light for a friend

You

¿Qué hackathons hay abiertos este mes? Usa DevRelay

Agent

▸ Skill {"skill": "devrelay-challenges", "args": "Hackathons abiertos en octubre 2026"}

Input

{
  "skill": "devrelay-challenges",
  "args": "Hackathons abiertos en octubre 2026"
}

Output

Launching skill: devrelay-challenges

▸ ToolSearch select:mcp__devrelay-gateway__get_challenges,mcp__devrelay-gateway__search_ml...

Input

select:mcp__devrelay-gateway__get_challenges,mcp__devrelay-gateway__search_mlh_events,mcp__devrelay-gateway__get_events

Output

▸ mcp__devrelay-gateway__get_challenges {}

Input

{
}

Output

{"Ok":[{"id":82,"title":"Hacktoberfest Open-Source AI Challenge: Week 4","slug":"hacktoberfest-week4-2026-10-26","description":"Register now to be notified as details drop.","details":null,"full_details":"","type_of":"challenge","starts_at":"2026-10-26T16:00:00.000Z","ends_at":"2026-10-31T06:59:00.000Z","published":true,"cover_image":"https://dev-to-uploads.s3.us-east-2.amazonaws.com/uploads/event
… [15668 more characters trimmed]

▸ mcp__devrelay-gateway__search_mlh_events {"event_format": "hackathon", "starts_before": "2026-10-31", "limit": 50}

Input

{
  "event_format": "hackathon",
  "starts_before": "2026-10-31",
  "limit": 50
}

Output

{"Ok":{"events":[{"id":"01a062af-7088-c530-82c5-b593e5825360","name":"FinHack$","slug":"finhack-33","status":"canceled","event_format":"hackathon","starts_at":1790903471,"ends_at":1791158399,"starts_at_utc":"2026-10-02T01:11:11Z","ends_at_utc":"2026-10-04T23:59:59Z","time_zone":null,"private":false,"website_url":null,"submission_url":null,"self_check_in_mode":"disabled","check_in":{"state":"event_
… [22540 more characters trimmed]

Esto es lo que hay en DevRelay para octubre de 2026 (hoy es 4 de octubre). Todas las horas están en UTC.

DEV Challenges (online)

Abiertos ahora
| Reto | Cierre | Queda |
|---|---|---|
| Hacktoberfest Weekend Challenge: Build for a Friend | 5 oct, 06:59 | ~1 día |
| Kaggle Benchmark Writing Challenge | 12 oct, 06:59 | ~8 días |

Build for a Friend: hay que construir algo con IA open-source como pieza central (un modelo open-weight, un framework open-source o inferencia local) que resuelva un problema real de una persona concreta. Tiene que ser un proyecto nuevo. Se entrega publicando un post en DEV con el tag #hf26challenge. Hay $2,450 en premios para 17 ganadores. Al jurado lo que más le pesa es la calidad del post. Hay categorías de patrocinador, entre ellas Gemma, DigitalOcean, Render, MongoDB Atlas y ElevenLabs.
Kaggle: el organizador todavía no ha publicado las reglas detalladas.

Próximos: la serie Hacktoberfest Open-Source AI Challenge sigue con el mismo prompt y un tema nuevo cada lunes. Cada semana es una entrega aparte:
- Semana 1: del 5 oct, 16:00, al 12 oct, 06:59
- Semana 2: del 12 al 18 oct
- Semana 3: del 19 al 25 oct
- Semana 4: del 26 al 31 oct

El Sanity Challenge ya cerró (4 oct, 06:59).

Hackathons MLH (presenciales; la inscripción sigue abierta)

Dejé fuera los cancelados y los de este fin de semana, que terminan hoy.

9–11 oct: HackNC, HackBIOS, Hack Knight, Hackathon TecNM Campus Saltillo (México), Hacklanta, Knight Hacks IX, HackRU, Palmetto Hacks, hackPHS, CodeRED Orion, Innohacks 4.0
16–18 oct: SwampHacks XII, Hack the Valley, HackGB, Girls Hoo Hack, DubHacks, IKU Womxn in STEM (Reino Unido), Technica, LA Hacks AI Hackathon
23–25 oct: Cal Hacks 13.0, SBUHacks, OxHack '26 (Oxford), HackTX 26, Hack at Penn State
31 oct – 1 nov: hackCBS 9.0 (India)

Tu carpeta se llama Hacktoberfest/2026/DEV Challenges/1, así que puede que ya estés con el Weekend Challenge. Si es así, cierra en menos de un día. Puedo revisar tu proyecto contra las reglas y dejarte el post de entrega como borrador. También puedo inscribirte en alguno de los hackathons MLH.

You

Efectivamente, quiero participar en Build for a Friend, pero no se ni por donde empezar, no se como hacer lo que hay que hacer

Agent

▸ Bash cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && ls -la && date -u; wh...

Input

$ cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && ls -la && date -u; which ollama node python git 2>/dev/null; ollama list 2>/dev/null

Output

Exit code 127
total 0
drwxr-xr-x 1 david 197609 0 Oct  4 10:30 .
drwxr-xr-x 1 david 197609 0 Oct  4 10:30 ..
Sun Oct  4 15:43:36 UTC 2026
~/AppData/Local/Microsoft/WindowsApps/python
/mingw64/bin/git

Tranquilo, es más sencillo de lo que parece. Te cuento lo que hay que hacer y el tiempo que queda.

Tiempo

Son las 15:43 UTC y la entrega cierra el 5 oct a las 06:59 UTC, así que quedan unas 15 horas. Da para algo pequeño, y el reto lo pide así: "No tiene que ser grande. Tiene que importarle a esa persona."

Lo que hay que entregar

Al final todo es un post en DEV con el tag #hf26challenge que incluya:
1. Qué construiste y para quién: una persona real (tu madre, un amigo, tu abuelo...).
2. Una demo: capturas, un GIF o un vídeo corto.
3. El código: un repo público en GitHub.
4. Por qué es importante que sea open-source: por ejemplo, que funciona sin internet, que los datos no salen de su ordenador o que no cuesta nada usarlo.

El jurado puntúa sobre todo la calidad del post. Una herramienta sencilla con una buena historia gana a un proyecto complejo mal contado.

El plan (yo me encargo de la parte técnica)

Tú eliges la persona y el problema. Es lo único que no puedo decidir yo.
Yo construyo con tu ayuda una app pequeña con un modelo abierto (por ejemplo Gemma con Ollama, que corre en tu PC). Usar Gemma además te hace optar a la categoría "Best Use of Gemma", de $200.
Tú la pruebas y, si puedes, se la enseñas a esa persona y apuntas qué te dice. Eso da puntos extra.
Yo te redacto el post como borrador en DEV y tú lo revisas y lo publicas.

Tu carpeta está vacía y no tienes Ollama instalado. Lo instalamos en el paso 2: tarda unos 10 minutos.

Lo que necesito de ti ahora

¿Para quién lo vas a construir y qué le cuesta en su día a día? Basta con una frase. Por ejemplo:
- "Mi madre recibe cartas del banco o de Hacienda y no las entiende": una app que se las explica en lenguaje sencillo, sin subir sus documentos a ningún sitio.
- "Mi abuelo se olvida de qué pastillas toma": un asistente que lee la caja del medicamento y le recuerda la pauta.
- "Un amigo está aprendiendo inglés": un compañero de conversación con paciencia infinita que funciona sin conexión.
- "Mi compañero de piso tiene alergias": un planificador de comidas que las tiene en cuenta.

Dime la tuya, aunque sea solo una idea a medias, y empezamos.

You

Para ese amigo que nunca llega a fin de més, una extensión de navegador o una aplicación que detecte que está en una web de tienda, y que indique de manera visual si el producto que está viendo se lo puede permitir o no. La aplicación pediría como configuración el income mensual, los gastos fijos mensuales y, si ahorra dinero cada més, para, en función del día del mes en el que estás, decirte si te lo deberías permitir o no. También decirte en cuantos meses podrías permitirtelo sin descapitalizarte. No se si se puede hacer una aplicación así con IA en local.

Agent

Sí, se puede hacer, y que la IA corra en local es justamente lo mejor de la idea para el post: ni lo que navega tu amigo ni sus datos de dinero salen nunca de su ordenador. Empiezo instalando Ollama, que es el programa que ejecuta el modelo de IA en local.

▸ PowerShell winget install --id Ollama.Ollama -e --accept-source-agreements --accept-pack...

Input

winget install --id Ollama.Ollama -e --accept-source-agreements --accept-package-agreements --silent

Output

Command running in background with ID: bn91zozdg. Output is being written to: <scratchpad> You will be notified when it completes. To check interim output, use Read on that file path.

▸ ToolSearch select:mcp__devrelay-gateway__search_dev_to_semantic,WebFetch

Input

select:mcp__devrelay-gateway__search_dev_to_semantic,WebFetch

Output

▸ WebFetch {"url": "https://ollama.com/library", "prompt": "List the Gemma model familie...

Input

{
  "url": "https://ollama.com/library",
  "prompt": "List the Gemma model families available (gemma, gemma2, gemma3, gemma4, etc.) with their tags/sizes, especially small ones (1b-4b) suitable for a laptop."
}

Output

# Gemma Model Families Available

Based on the Ollama library, here are the Gemma model families suitable for laptops:

**Gemma** (Original)
- Sizes: 2B, 7B
- "Gemma is a family of lightweight, state-of-the-art open models built by Google DeepMind"

**Gemma2**
- Sizes: 2B, 9B, 27B
- "Google Gemma 2 is a high-performing and efficient model available in three sizes"

**Gemma3**
- Sizes: 270M, 1B, 4B
… [544 more characters trimmed]

▸ mcp__devrelay-gateway__search_dev_to_semantic Chrome extension calling local Ollama API CORS OLLAMA_ORIGINS

Input

Chrome extension calling local Ollama API CORS OLLAMA_ORIGINS

Output

{"Ok":[{"id":4573937,"title":"Ollama API: A Practical Guide with Examples","description":"A practical Ollama API guide: generate and chat endpoints, streaming, options, structured JSON, tool calling, Python, and the OpenAI-compatible route.","tags":["ollama","api","llm","python"],"path":"/amareswer/ollama-api-a-practical-guide-with-examples-4di9","url":"https://dev.to/amareswer/ollama-api-a-practi
… [1832 more characters trimmed]

Gemma 3 está disponible en Ollama. Para un portátil la mejor opción es gemma3:4b, que además te hace optar al premio "Best Use of Gemma". Mientras se instala Ollama, escribo la extensión.

▸ Bash cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && git init -q && mkdir ...

Input

$ cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && git init -q && mkdir -p extension/icons && powershell -NoProfile -Command "Get-CimInstance Win32_ComputerSystem | Select TotalPhysicalMemory; Get-CimInstance Win32_VideoController | Select Name,AdapterRAM"

Output

TotalPhysicalMemory
-------------------
        17095188480

▸ Write ~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\manifest.json { ...

Input

~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\manifest.json

{
  "manifest_version": 3,
  "name": "FinDeMes",
  "version": "0.1.0",
  "description": "¿Me lo puedo permitir? Un semáforo para cada producto que miras, con IA que corre en tu ordenador.",
  "permissions": ["storage"],
  "host_permissions": ["http://localhost:11434/*", "http://127.0.0.1:11434/*"],
  "background": {
    "serv
… [433 more characters trimmed]

Output

File created successfully at: ~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\manifest.json (file state is current in your context — no need to Read it back)

▸ Write ~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\budget.js // Lógic...

Input

~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\budget.js

// Lógica del presupuesto. Sin IA: las cuentas tienen que ser exactas y explicables.

export const DEFAULT_SETTINGS = {
  name: "",
  income: 0,        // ingresos netos al mes
  fixed: 0,         // gastos fijos al mes (alquiler, facturas, suscripciones...)
  savings: 0,       // lo que aparta cada mes para ahorrar
  payday: 1, 
… [2835 more characters trimmed]

Output

File created successfully at: ~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\budget.js (file state is current in your context — no need to Read it back)

▸ Write ~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\background.js impo...

Input

~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\background.js

import { DEFAULT_SETTINGS, evaluate, money } from "./budget.js";

const EXTRACT_SCHEMA = {
  type: "object",
  properties: {
    is_product_page: { type: "boolean" },
    product_name: { type: "string" },
    price: { type: "number" },
    currency: { type: "string" },
  },
  required: ["is_product_page", "product_name", "pri
… [6143 more characters trimmed]

Output

File created successfully at: ~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\background.js (file state is current in your context — no need to Read it back)

▸ Write ~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\content.js // Se e...

Input

~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\content.js

// Se ejecuta en cada página. Solo despierta a la IA si la página "huele" a tienda.
(() => {
  const BUY_WORDS = /añadir al carrito|añadir a la cesta|comprar ahora|add to (cart|basket|bag)|buy now|in den warenkorb|ajouter au panier|aggiungi al carrello|adicionar ao carrinho/i;
  const PRICE_RE = /(\d{1,3}([.,\s]\d{3})*([.,]\d{2}
… [7061 more characters trimmed]

Output

File created successfully at: ~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\content.js (file state is current in your context — no need to Read it back)

▸ Write ~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\options.html <!doc...

Input

~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\options.html

<!doctype html>
<html lang="es">
<head>
  <meta charset="utf-8">
  <title>FinDeMes · Configuración</title>
  <style>
    body { font: 15px/1.5 system-ui, sans-serif; max-width: 520px; margin: 40px auto; padding: 0 16px; color: #1a202c; }
    h1 { margin-bottom: 4px; }
    .sub { color: #4a5568; margin-top: 0; }
    label { dis
… [2081 more characters trimmed]

Output

File created successfully at: ~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\options.html (file state is current in your context — no need to Read it back)

▸ Write ~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\options.js import ...

Input

~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\options.js

import { DEFAULT_SETTINGS, money } from "./budget.js";

const FIELDS = ["name", "income", "fixed", "savings", "payday", "currency", "model", "ollamaUrl"];
const NUMERIC = new Set(["income", "fixed", "savings", "payday"]);
const $ = (id) => document.getElementById(id);

function read() {
  const s = {};
  for (const f of FIELDS) 
… [1471 more characters trimmed]

Output

File created successfully at: ~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\options.js (file state is current in your context — no need to Read it back)

▸ Write ~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\popup.html <!docty...

Input

~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\popup.html

<!doctype html>
<html lang="es">
<head>
  <meta charset="utf-8">
  <style>
    body { font: 14px/1.4 system-ui, sans-serif; width: 300px; margin: 0; padding: 14px; color: #1a202c; }
    h2 { margin: 0 0 10px; font-size: 16px; }
    .meter { height: 10px; background: #e2e8f0; border-radius: 5px; overflow: hidden; margin: 8px 0; }
… [535 more characters trimmed]

Output

File created successfully at: ~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\popup.html (file state is current in your context — no need to Read it back)

▸ Write ~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\popup.js import { ...

Input

~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\popup.js

import { DEFAULT_SETTINGS, monthSummary, money } from "./budget.js";

const $ = (id) => document.getElementById(id);
$("options").onclick = () => chrome.runtime.openOptionsPage();

async function render() {
  const stored = await chrome.storage.local.get(["settings", "purchases"]);
  const settings = { ...DEFAULT_SETTINGS, ...stor
… [1569 more characters trimmed]

Output

File created successfully at: ~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\popup.js (file state is current in your context — no need to Read it back)

▸ Bash cat "<scratchpad>" | tail -5; ls "%LOCALAPPDATA%/Programs/Ollama" 2>/dev/null

Input

$ cat "<scratchpad>" | tail -5; ls "%LOCALAPPDATA%/Programs/Ollama" 2>/dev/null

Output

El hash del instalador se verificó correctamente
Iniciando instalación de paquete...
Instalado correctamente

[exited with code 0]
app.ico
lib
ollama app.exe
ollama.exe
unins000.dat
unins000.exe
unins000.msg

Ollama ya está instalado. Ahora descargo el modelo Gemma, que pesa unos 3 GB.

▸ Bash O="%LOCALAPPDATA%/Programs/Ollama/ollama.exe"; (curl -s localhost:11434/api/v...

Input

$ O="%LOCALAPPDATA%/Programs/Ollama/ollama.exe"; (curl -s localhost:11434/api/version || (nohup "$O" serve >/dev/null 2>&1 & sleep 4; curl -s localhost:11434/api/version)); echo; "$O" pull gemma3:4b 2>&1 | tail -2

Output

Command running in background with ID: bml7x7e0e. Output is being written to: <scratchpad> You will be notified when it completes. To check interim output, use Read on that file path.

▸ PowerShell Add-Type -AssemblyName System.Drawing $dir = "~\Documents\Hacktoberfest\2026\...

Input

Add-Type -AssemblyName System.Drawing
$dir = "~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\icons"
foreach ($s in 16,48,128) {
  $bmp = New-Object System.Drawing.Bitmap $s,$s
  $g = [System.Drawing.Graphics]::FromImage($bmp)
  $g.SmoothingMode = 'AntiAlias'
  $g.Clear([System.Drawing.Color]::Transparent)
  $w = $s/3
  $cols = '#e53e3e','#d69e2e','#1f9d55'
  for ($i=0; $i -lt 3; $i++) {
… [659 more characters trimmed]

Output

Name        Length
----        ------
icon128.png   1676
icon16.png     332
icon48.png     672
C:\Program Files\Google\Chrome\Application\chrome.exe
C:\Program Files (x86)\Microsoft\Edge\Application\msedge.exe

▸ Bash S="<scratchpad>"; mkdir -p "$S"; cp "~/Documents/Hacktoberfest/2026/DEV Chall...

Input

$ S="<scratchpad>"; mkdir -p "$S"; cp "~/Documents/Hacktoberfest/2026/DEV Challenges/1/extension/budget.js" "$S/"; cat > "$S/t.html" <<'EOF'
<pre id=o></pre><script type=module>
import {evaluate,payCycle} from "./budget.js";
const st={income:1500,fixed:900,savings:100,payday:1,currency:"EUR"}; // libre 500
const out=[];const T=(l,p,pur,d)=>{const r=evaluate(p,st,pur,new Date(d));out.push(l+" "+JSON.
… [957 more characters trimmed]

Output

dia3 30€ {"v":"green","el":3,"days":31,"left":29,"onPace":48.39,"rem":500}
dia3 200€ {"v":"yellow","el":3,"days":31,"left":29,"onPace":48.39,"rem":500,"wait":10}
dia28 200€ {"v":"green","el":28,"days":31,"left":4,"onPace":451.61,"rem":500}
dia15 900€ {"v":"red","el":15,"days":31,"left":17,"onPace":241.94,"rem":500,"mA":1,"mH":2}
dia15 gastado450 100€ {"v":"red","el":15,"days":31,"left":17,"onPace"
… [1154 more characters trimmed]

Las cuentas del presupuesto dan lo esperado en todos los casos que he probado. Veamos cómo va la descarga del modelo.

▸ Bash cat "<scratchpad>" | tr '\r' '\n' | tail -3

Input

$ cat "<scratchpad>" | tr '\r' '\n' | tail -3

Output

{
  "version": "0.35.1"
}

▸ Bash curl -s -o /dev/null -w "%{http_code}\n" -H "Origin: chrome-extension://abcde...

Input

$ curl -s -o /dev/null -w "%{http_code}\n" -H "Origin: chrome-extension://abcdefghijklmnop" localhost:11434/api/tags; "%LOCALAPPDATA%/Programs/Ollama/ollama.exe" list

Output

403
NAME    ID    SIZE    MODIFIED

Ollama rechaza las peticiones que vienen de extensiones del navegador. Para que la extensión pueda hablar con él hay que darle permiso con una variable de entorno:

▸ PowerShell [Environment]::SetEnvironmentVariable("OLLAMA_ORIGINS", "chrome-extension://*...

Input

[Environment]::SetEnvironmentVariable("OLLAMA_ORIGINS", "chrome-extension://*", "User"); [Environment]::GetEnvironmentVariable("OLLAMA_ORIGINS","User")

Output

chrome-extension://*

Mientras termina la descarga, escribo el README y la licencia, que el reto pide que el código sea abierto:

▸ Write ~\Documents\Hacktoberfest\2026\DEV Challenges\1\README.md # FinDeMes 💸 **¿M...

Input

~\Documents\Hacktoberfest\2026\DEV Challenges\1\README.md

# FinDeMes 💸

**¿Me lo puedo permitir?** Una extensión de navegador que, cuando estás mirando un producto en cualquier tienda online, te enseña un semáforo:

- 🟢 **Te lo puedes permitir**: entra en tu ritmo de gasto del mes.
- 🟡 **Puedes, pero te aprieta**: te dice cuánto te quedaría al día hasta que cobres y cuántos días esperar para que 
… [2510 more characters trimmed]

Output

File created successfully at: ~\Documents\Hacktoberfest\2026\DEV Challenges\1\README.md (file state is current in your context — no need to Read it back)

▸ Bash cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && cat > LICENSE <<'EOF'...

Input

$ cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && cat > LICENSE <<'EOF'
MIT License

Copyright (c) 2026 David Antón

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distrib
… [793 more characters trimmed]

Output

David Anton Alvarez
<email>

▸ Bash cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && sed -i 's/Copyright (...

Input

$ cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && sed -i 's/Copyright (c) 2026 David Antón/Copyright (c) 2026 David Anton Alvarez/' LICENSE; "%LOCALAPPDATA%/Programs/Ollama/ollama.exe" list; cat "<scratchpad>" | tr '\r' '\n' | tail -2

Output

NAME    ID    SIZE    MODIFIED 
{"version":"0.35.1"}

▸ Monitor until "%LOCALAPPDATA%/Programs/Ollama/ollama.exe" list | grep -q gemma3; do s...

Input

until "%LOCALAPPDATA%/Programs/Ollama/ollama.exe" list | grep -q gemma3; do sleep 15; done; echo listo

Output

<tool_use_error>InputValidationError: Monitor failed due to the following issue:
The required parameter `description` is missing

This tool's schema was not sent to the API — it was not in the discovered-tool set derived from message history. Without the schema in your prompt, typed parameters (arrays, numbers, booleans) get emitted as strings and the client-side parser rejects them. Load the tool
… [1178 more characters trimmed]

▸ ToolSearch select:Monitor

Input

select:Monitor

Output

▸ Bash python --version 2>&1; py --version 2>&1

Input

$ python --version 2>&1; py --version 2>&1

Output

Python 3.14.0b2
Python 3.14.0b2

▸ Write ~\Documents\Hacktoberfest\2026\DEV Challenges\1\demo\index.html <!doctype ht...

Input

~\Documents\Hacktoberfest\2026\DEV Challenges\1\demo\index.html

<!doctype html>
<html lang="es">
<head>
  <meta charset="utf-8">
  <title>Auriculares SonoMax Pro – TiendaDemo</title>
  <style>
    body { font: 16px/1.5 system-ui, sans-serif; margin: 0; color: #222; }
    header { background: #222; color: #fff; padding: 12px 24px; }
    main { display: flex; gap: 32px; max-width: 960px; margin: 32
… [1135 more characters trimmed]

Output

File created successfully at: ~\Documents\Hacktoberfest\2026\DEV Challenges\1\demo\index.html (file state is current in your context — no need to Read it back)

▸ Bash until "%LOCALAPPDATA%/Programs/Ollama/ollama.exe" list | grep -q gemma3; do s...

Input

$ until "%LOCALAPPDATA%/Programs/Ollama/ollama.exe" list | grep -q gemma3; do sleep 10; done; "%LOCALAPPDATA%/Programs/Ollama/ollama.exe" list

Output

Command did not complete within its 600s timeout and was moved to the background (ID: blahiryga). Output is being written to: <scratchpad> You will be notified when it completes. To check interim output, use Read on that file path.

▸ Bash cat "<scratchpad>" | tr '\r' '\n' | grep -v '^$' | tail -3; "%LOCALAPPDATA%/P...

Input

$ cat "<scratchpad>" | tr '\r' '\n' | grep -v '^$' | tail -3; "%LOCALAPPDATA%/Programs/Ollama/ollama.exe" list

Output

pulling aeda25e63ebd:   6% ▕█                 ▏ 210 MB/3.3 GB                  [K[?25h[?2026l
Error: max retries exceeded: write ~\.ollama\models\blobs\sha256-aeda25e63ebd698fab8638ffb778e68bed908b960d39d0becc650fa981609d25-partial: There is not enough space on the disk.
[exited with code 0]
NAME    ID    SIZE    MODIFIED

La descarga ha fallado porque no queda espacio en el disco. Miro cuánto espacio libre hay:

▸ PowerShell Get-PSDrive -PSProvider FileSystem | Select-Object Name, @{n='LibreGB';e={[ma...

Input

Get-PSDrive -PSProvider FileSystem | Select-Object Name, @{n='LibreGB';e={[math]::Round($_.Free/1GB,2)}}, @{n='UsadoGB';e={[math]::Round($_.Used/1GB,1)}}; $p = "$env:USERPROFILE\.ollama\models\blobs"; if (Test-Path $p) { Get-ChildItem $p | Select Name, @{n='MB';e={[math]::Round($_.Length/1MB)}} }

Output

Name                                                                               LibreGB UsadoGB
----                                                                               ------- -------
C                                                                                     0,05   475,3
D                                                                                    39,33   696,9

… [1781 more characters trimmed]

▸ TaskStop {"task_id": "blahiryga"}

Input

{
  "task_id": "blahiryga"
}

Output

{"message":"Successfully stopped task: blahiryga (until \"%LOCALAPPDATA%/Programs/Ollama/ollama.exe\" list | grep -q gemma3; do sleep 10; done; \"%LOCALAPPDATA%/Programs/Ollama/ollama.exe\" list)","task_id":"blahiryga","task_type":"local_bash","command":"until \"%LOCALAPPDATA%/Programs/Ollama/ollama.exe\" list | grep -q gemma3; do sleep 10; done; \"%LOCALAPPDATA%/Programs/Ollama/ollama.exe\" list"
… [1 more characters trimmed]

▸ PowerShell Get-Process ollama*, "ollama app" -ErrorAction SilentlyContinue | Stop-Proces...

Input

Get-Process ollama*, "ollama app" -ErrorAction SilentlyContinue | Stop-Process -Force -Confirm:$false
Remove-Item "$env:USERPROFILE\.ollama\models\blobs\*-partial*" -Force -Confirm:$false
New-Item -ItemType Directory -Force "D:\ollama\models" | Out-Null
[Environment]::SetEnvironmentVariable("OLLAMA_MODELS", "D:\ollama\models", "User")
$env:OLLAMA_MODELS = "D:\ollama\models"; $env:OLLAMA_ORIGINS = 
… [333 more characters trimmed]

Output

origin check: 200
C_LibreGB
---------
     1,96

Ollama ya acepta peticiones de la extensión y C: ha recuperado 2 GB. Descargo de nuevo el modelo, esta vez en D:.

▸ Bash "%LOCALAPPDATA%/Programs/Ollama/ollama.exe" pull gemma3:4b 2>&1 | tr '\r' '\n...

Input

$ "%LOCALAPPDATA%/Programs/Ollama/ollama.exe" pull gemma3:4b 2>&1 | tr '\r' '\n' | grep -E "success|Error" ; "%LOCALAPPDATA%/Programs/Ollama/ollama.exe" list

Output

Command running in background with ID: bjuggsm9z. Output is being written to: <scratchpad> You will be notified when it completes. To check interim output, use Read on that file path.

▸ Write <scratchpad> # Reproduce la llamada de extracción de background.js contra va...

Input

<scratchpad>

# Reproduce la llamada de extracción de background.js contra varias páginas de texto.
import json, re, time, urllib.request

bg = open(r"~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\background.js", encoding="utf-8").read()
prompt = re.search(r"const EXTRACT_PROMPT = `(.*?)`;", bg, re.S).group(1)
schema = {"type": "object", "properties": {"is_product_page": {"type": "bool
… [1619 more characters trimmed]

Output

File created successfully at: <scratchpad> (file state is current in your context — no need to Read it back)

▸ Bash until grep -qE "success|Error" "<scratchpad>"; do sleep 10; done; cat "<scrat...

Input

$ until grep -qE "success|Error" "<scratchpad>"; do sleep 10; done; cat "<scratchpad>"

Output

success [K[?25h[?2026l
NAME         ID              SIZE      MODIFIED               
gemma3:4b    a2af6cc3eb7f    3.3 GB    Less than a second ago    

[exited with code 0]

Gemma ya está descargado. Pruebo si sabe leer bien las páginas:

▸ Bash PYTHONIOENCODING=utf-8 python "<scratchpad>"

Input

$ PYTHONIOENCODING=utf-8 python "<scratchpad>"

Output

demo      92.9s {"is_product_page":true, "product_name":"SonoMax Pro", "price":179.99, "currency":"EUR"}
listado    8.1s {"is_product_page":false,"product_name":"zapatillas","price":129.95,"currency":"EUR"}
en         9.4s {"is_product_page": true, "product_name": "Acer Swift Go 14 Laptop", "price": 749.00, "currency": "USD"}

▸ Edit ~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\background.js --- o...

Input

~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\background.js
--- old
chrome.runtime.onInstalled.addListener(({ reason }) => {
  if (reason === "install") chrome.runtime.openOptionsPage();
});
+++ new
// Cargar el modelo en memoria al abrir el navegador: la primera respuesta en frío puede tardar más de un minuto.
async function warmUp() {
  const settings = await getSettings();
  fetch(`
… [336 more characters trimmed]

Output

The file ~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\background.js has been updated successfully. (file state is current in your context — no need to Read it back)

▸ Bash cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && sed -i 's/keep_alive:...

Input

$ cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && sed -i 's/keep_alive: "15m"/keep_alive: "30m"/' extension/background.js && cat > "$TEMP/note.json" <<'EOF'
{"model":"gemma3:4b","stream":false,"options":{"temperature":0.7},"messages":[{"role":"system","content":"Eres un amigo cercano, sincero y con humor que ayuda a llegar a fin de mes. Hablas con Pablo. Escribe UNA sola frase en español (má
… [436 more characters trimmed]

Output

Oye, con 11 euros al día y un SonoMax Pro que te aprieta, ¡es el momento de priorizar tu paz mental, ¿no?


real    0m6.356s
user    0m0.015s
sys 0m0.046s

La frase del amigo sale en unos 6 s. Guardo el trabajo en git:

▸ Bash cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && git add -A && git com...

Input

$ cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && git add -A && git commit -q -m "FinDeMes: extensión que dice si te puedes permitir un producto, con Gemma en local

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>" && git log --oneline && git status --short

Output

warning: in the working copy of 'LICENSE', LF will be replaced by CRLF the next time Git touches it
warning: in the working copy of 'README.md', LF will be replaced by CRLF the next time Git touches it
warning: in the working copy of 'demo/index.html', LF will be replaced by CRLF the next time Git touches it
warning: in the working copy of 'extension/background.js', LF will be replaced by CRLF the
… [909 more characters trimmed]

Y arranco la tienda de prueba para que puedas verla funcionar ya mismo:

▸ Bash cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1/demo" && python -m http.s...

Input

$ cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1/demo" && python -m http.server 8000

Output

Command running in background with ID: bochs308p. Output is being written to: <scratchpad> You will be notified when it completes. To check interim output, use Read on that file path.
Session cwd remains ~\Documents\Hacktoberfest\2026\DEV Challenges\1; directory changes made by the backgrounded command do not apply to subsequent commands.

Sí se podía hacer con IA en local, y FinDeMes ya está construido. Es una extensión para Chrome o Edge que, al abrir la ficha de un producto en cualquier tienda, enseña un semáforo en la esquina:

🟢 Te lo puedes permitir: dice cuánto le quedaría al día hasta cobrar.
🟡 Puedes, pero te aprieta: además dice cuántos días esperar para que entre en su ritmo de gasto.
🔴 Ahora no: dice en cuántos meses podría comprarlo sin tocar sus ahorros.

Lleva también una frase corta escrita por Gemma, como lo diría un amigo, y un botón "Lo he comprado" que resta el gasto del mes. Le he añadido el día de cobro, porque el mes de tu amigo empieza cuando cobra, no el día 1.

El reparto de trabajo, que es lo que contarás en el post: Gemma lee la página, decide si es un producto y saca el precio real, sin programar un lector para cada tienda. Las cuentas del presupuesto son código normal, porque tienen que ser exactas.

Lo que he comprobado:
- Las cuentas del presupuesto dan lo esperado en todos los casos que probé.
- Gemma acierta en tres páginas de prueba: elige el precio bueno y no el tachado, el del envío ni el de plazos; distingue un listado de búsqueda de una ficha de producto; y funciona en inglés.
- La frase del amigo tarda unos 6 s y el semáforo unos 9 s. La primera vez que se carga el modelo tarda más de un minuto; por eso la extensión lo precarga al abrir el navegador.

Lo que no he probado: la extensión cargada en el navegador de principio a fin, porque Chrome no me deja instalarla por ti. Si algo falla al probarla, pégame el error.

Dos cosas que he cambiado en tu PC:
- Tu disco C: estaba lleno (le quedaban 50 MB) y la primera descarga del modelo falló. He movido los modelos de Ollama a D:\ollama\models y C: tiene ahora unos 2 GB libres. Te conviene liberar más espacio pronto, porque Windows va mal con el disco tan lleno.
- He añadido una variable de entorno para que Ollama acepte peticiones de extensiones del navegador.

Pruébala ahora (5 minutos)

En Chrome abre chrome://extensions, activa Modo desarrollador, pulsa Cargar descomprimida y elige la carpeta extension.
Se abre la configuración. Mete los números de tu amigo (o unos inventados) y pulsa Probar conexión en "IA local", que debería decir "Todo listo ✔".
Abre la tienda de prueba que tengo arrancada en http://localhost:8000 y después prueba en una tienda real.

Lo que queda (cierra en ~13 h, el 5 oct a las 06:59 UTC; en España, a las 08:59)

Que la pruebe tu amigo y apuntar qué dice. Da puntos extra y es lo mejor del post.
Capturas o un GIF del semáforo en los tres colores.
Subir el código a GitHub. Ya está guardado en git en local. Si quieres, creo el repo público y lo subo, pero confírmamelo antes porque lo publica.
El post en DEV. Te lo redacto como borrador con el tag #hf26challenge, centrado en tu amigo y en por qué es importante que todo sea local: sus webs y su dinero no salen de su ordenador. Tú lo revisas y lo publicas.

¿Lo pruebas y me cuentas qué tal?

You

Antes de hacer nada, quiero dos cosas:

Agent

Vale, no toco nada. Tu mensaje se ha cortado después de "dos cosas:" y no me ha llegado cuáles son. ¿Me las escribes?

You

Antes de hacer nada, quiero dos cosas:

[Request interrupted by user]

Antes de hacer nada quiero dos cosas: la primera, es que el readme lo quiero en inglés primero (el organizador del hacktoberfest es angloparlante) y después en Español. Por esta misma razón, hay que cambiar el nombre de la app, que quiero que sea Should I Buy, y que los textos de meter los datos estén también en inglés

Agent

▸ Grep {"pattern": "FinDeMes|findemes", "path": "~\\Documents\\Hacktoberfest\\2026\\...

Input

{
  "pattern": "FinDeMes|findemes",
  "path": "~\\Documents\\Hacktoberfest\\2026\\DEV Challenges\\1",
  "output_mode": "content",
  "-i": true
}

Output

extension\content.js:105:      title.textContent = "FinDeMes no puede pensar";
extension\options.html:5:  <title>FinDeMes · Configuración</title>
extension\options.html:21:  <h1>FinDeMes 💸</h1>
extension\popup.html:18:  <h2>FinDeMes 💸</h2>
README.md:1:# FinDeMes 💸
README.md:13:FinDeMes ve **todas las tiendas que visitas** y conoce **tu sueldo, tus gastos y tus ahorros**. Esos datos no deberían sal
… [126 more characters trimmed]

▸ Bash cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1/extension" && sed -i 's/"...

Input

$ cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1/extension" && sed -i 's/"name": "FinDeMes"/"name": "Should I Buy"/; s/"default_title": "FinDeMes"/"default_title": "Should I Buy"/; s/"description": ".*"/"description": "Can I afford it? A traffic light for every product you look at, powered by AI that runs on your own computer."/' manifest.json && sed -i 's/FinDeMes no puede pensar/Should I Buy 
… [134 more characters trimmed]

Output

3:  "name": "Should I Buy",
5:  "description": "Can I afford it? A traffic light for every product you look at, powered by AI that runs on your own computer.",
21:    "default_title": "Should I Buy"

Ahora la página de configuración en inglés:

▸ Write ~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\options.html <!doc...

Input

~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\options.html

<!doctype html>
<html lang="en">
<head>
  <meta charset="utf-8">
  <title>Should I Buy · Settings</title>
  <style>
    body { font: 15px/1.5 system-ui, sans-serif; max-width: 520px; margin: 40px auto; padding: 0 16px; color: #1a202c; }
    h1 { margin-bottom: 4px; }
    .sub { color: #4a5568; margin-top: 0; }
    label { disp
… [2074 more characters trimmed]

Output

The file ~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\options.html has been updated successfully. (file state is current in your context — no need to Read it back)

▸ Bash cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1/extension" && sed -i \ -...

Input

$ cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1/extension" && sed -i \
 -e 's/`Te quedan ${money(free, s.currency)} libres al mes para el día a día y los caprichos.`/`You have ${money(free, s.currency)} free each month for day-to-day spending and treats.`/' \
 -e 's/"Con estos números no te queda dinero libre al mes."/"With these numbers you have no free money left each month."/' \
 -e 's/"Gua
… [361 more characters trimmed]

Output

17:  $("free").textContent =
32:  $("status").textContent = "Saved ✔";
33:  setTimeout(() => ($("status").textContent = ""), 2000);
38:  $("ollama").textContent = "Testing…";
41:    $("ollama").textContent = `Cannot connect: ${res.error}`;
43:    $("ollama").textContent = `Connected, but the model is missing. Run: ollama pull ${read().model}`;
45:    $("ollama").textContent = "All set ✔";

Ahora el README, primero en inglés y después en español:

▸ Write ~\Documents\Hacktoberfest\2026\DEV Challenges\1\README.md # Should I Buy 💸 ...

Input

~\Documents\Hacktoberfest\2026\DEV Challenges\1\README.md

# Should I Buy 💸

[English](#english) · [Español](#español)

## English

**Can I afford it?** A browser extension that shows you a traffic light whenever you're looking at a product in any online shop:

- 🟢 **You can afford it**: it fits your spending pace for the month.
- 🟡 **You can, but it's tight**: it tells you how much you'd have lef
… [5060 more characters trimmed]

Output

The file ~\Documents\Hacktoberfest\2026\DEV Challenges\1\README.md has been updated successfully. (file state is current in your context — no need to Read it back)

▸ Bash cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && grep -rni findemes . ...

Input

$ cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && grep -rni findemes . --exclude-dir=.git; python -c "import json;json.load(open('extension/manifest.json',encoding='utf-8'))" && git add -A && git commit -q -m "Rename to Should I Buy; English settings page and bilingual README

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>" && git log --oneline | head -1

Output

warning: in the working copy of 'README.md', LF will be replaced by CRLF the next time Git touches it
warning: in the working copy of 'extension/content.js', LF will be replaced by CRLF the next time Git touches it
warning: in the working copy of 'extension/manifest.json', LF will be replaced by CRLF the next time Git touches it
warning: in the working copy of 'extension/options.html', LF will be 
… [346 more characters trimmed]

He hecho los dos cambios y los he guardado en git:

README bilingüe: ahora va primero en inglés y después en español, con enlaces arriba para saltar a cada idioma.
La app se llama Should I Buy: he cambiado el nombre en el manifest, en la descripción de la extensión (ahora en inglés), en el popup y en el aviso de error. Toda la página de configuración está en inglés: los campos, las ayudas y los mensajes de guardar y de probar la conexión.

Si ya habías cargado la extensión en Chrome, pulsa el botón de recargar en chrome://extensions para ver los cambios.

Lo que sigue en español:
- El semáforo que aparece en las tiendas: "Te lo puedes permitir", el botón "Lo he comprado", etc.
- El popup con el resumen del mes.
- La frase que escribe Gemma.
- Las cantidades se escriben al estilo español ("179,99 €").

Si tu amigo habla español puede tener sentido dejarlo así, pero los jueces verán las capturas con textos en español. ¿Lo paso todo a inglés o lo dejo como está?

You

Pásalo todo al inglés, pero deja los presupuestos en Euros

Agent

▸ Read ~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\background.js

Input

~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\background.js

Output

1    import { DEFAULT_SETTINGS, evaluate, money } from "./budget.js";
2   
3   const EXTRACT_SCHEMA = {
4     type: "object",
5     properties: {
6       is_product_page: { type: "boolean" },
7       product_name: { type: "string" },
8       price: { type: "number" },
9       currency: { type: "string" },
10    },
11    required: ["is_product_page", "product_name", "price", "currency"],
12  };
13  
14  const EXTRACT_PROM
… [7053 more characters trimmed]

▸ Write ~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\background.js impo...

Input

~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\background.js

import { DEFAULT_SETTINGS, evaluate, money } from "./budget.js";

const EXTRACT_SCHEMA = {
  type: "object",
  properties: {
    is_product_page: { type: "boolean" },
    product_name: { type: "string" },
    price: { type: "number" },
    currency: { type: "string" },
  },
  required: ["is_product_page", "product_name", "pri
… [6470 more characters trimmed]

Output

The file ~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension\background.js has been updated successfully. (file state is current in your context — no need to Read it back)

▸ Bash cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1/extension" && sed -i \ -...

Input

$ cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1/extension" && sed -i \
 -e 's|// Lógica del presupuesto. Sin IA: las cuentas tienen que ser exactas y explicables.|// Budget logic. No AI here: the maths has to be exact and explainable.|' \
 -e 's|// ingresos netos al mes|// monthly net income|' \
 -e 's|// gastos fijos al mes (alquiler, facturas, suscripciones...)|// monthly fixed expenses (ren
… [1169 more characters trimmed]

Output

1:// Budget logic. No AI here: the maths has to be exact and explainable.
5:  income: 0,        // monthly net income
6:  fixed: 0,         // monthly fixed expenses (rent, bills, subscriptions...)
7:  savings: 0,       // what gets put aside every month
8:  payday: 1,        // day of the month the salary arrives (1-28)
11:  ollamaUrl: "http://localhost:11434",
20:// Current pay cycle: from the l
… [547 more characters trimmed]

Ahora el semáforo de las tiendas y el popup:

▸ Bash cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1/extension" && sed -i \ -...

Input

$ cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1/extension" && sed -i \
 -e 's|// Se ejecuta en cada página. Solo despierta a la IA si la página "huele" a tienda.|// Runs on every page. Only wakes the AI up if the page "smells" like a shop.|' \
 -e 's|// ---------- Interfaz ----------|// ---------- UI ----------|' \
 -e 's|green: "Te lo puedes permitir", yellow: "Puedes, pero te aprieta", red: 
… [1926 more characters trimmed]

Output

Exit code 2
grep: -P supports only unibyte and UTF-8 locales

▸ Grep {"pattern": "[áéíóúñ¿¡]|Pensando|Ahora no|carrito", "path": "~\\Documents\\Ha...

Input

{
  "pattern": "[áéíóúñ¿¡]|Pensando|Ahora no|carrito",
  "path": "~\\Documents\\Hacktoberfest\\2026\\DEV Challenges\\1\\extension",
  "output_mode": "content"
}

Output

content.js:3:  const BUY_WORDS = /añadir al carrito|añadir a la cesta|comprar ahora|add to (cart|basket|bag)|buy now|in den warenkorb|ajouter au panier|aggiungi al carrello|adicionar ao carrinho/i;

▸ Bash cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1/demo" && sed -i \ -e 's|...

Input

$ cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1/demo" && sed -i \
 -e 's|lang="es"|lang="en"|' \
 -e 's|Auriculares SonoMax Pro – TiendaDemo|SonoMax Pro Headphones – DemoShop|' \
 -e 's|TiendaDemo · Carrito (0)|DemoShop · Cart (0)|' \
 -e 's|Auriculares inalámbricos SonoMax Pro|SonoMax Pro Wireless Headphones|' \
 -e 's|1.284 valoraciones|1,284 ratings|' \
 -e 's|249,99 €|€249.99|' -e 's|179,9
… [468 more characters trimmed]

Output

13:    button { padding: 12px 24px; font-size: 16px; background: #f6ad55; border: 0; border-radius: 8px; }
22:      <h1>SonoMax Pro Wireless Headphones</h1>
24:      <p><span class="old">€249.99</span> −28%</p>
25:      <p class="price">€179.99</p>
26:      <p>Shipping: €4.95 · Free over €199</p>
27:      <p>Or 3 instalments of €60.00</p>
28:      <button>Add to cart</button>
33:    <p>SonoMax Cas
… [64 more characters trimmed]

Actualizo las menciones al botón en el README y vuelvo a probar a Gemma con las instrucciones ya en inglés:

▸ Bash cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && sed -i -e 's|The \*\*...

Input

$ cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && sed -i -e 's|The \*\*Lo he comprado\*\* ("I bought it") button|The **I bought it** button|' -e 's|Con el botón \*\*Lo he comprado\*\* apuntas|Con el botón **I bought it** apuntas|' README.md && grep -n "bought" README.md
S="<scratchpad>"
sed -i 's|"demo": "TiendaDemo.*",$|"demo": "DemoShop · Cart (0) 🎧 SonoMax Pro Wireless Headphones ⭐⭐⭐⭐☆ 1,
… [989 more characters trimmed]

Output

37:   - The **I bought it** button logs the purchase and subtracts it from the month.
83:   - Con el botón **I bought it** apuntas el gasto y se resta del mes.
demo      28.5s {"is_product_page":true, "product_name":"SonoMax Pro Wireless Headphones", "price":179.99, "currency":"EUR"}
listado    8.8s {"is_product_page":false,"product_name":"","price":0,"currency":""}
en        11.9s {"is_product_
… [205 more characters trimmed]

Con las instrucciones en inglés Gemma sigue acertando en los tres casos. Guardo:

▸ Bash cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && git add -A && git com...

Input

$ cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && git add -A && git commit -q -m "Translate the whole extension and demo shop to English; keep budgets in euros

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>" 2>/dev/null; git log --oneline | head -1; git status --short

Output

warning: in the working copy of 'README.md', LF will be replaced by CRLF the next time Git touches it
warning: in the working copy of 'demo/index.html', LF will be replaced by CRLF the next time Git touches it
warning: in the working copy of 'extension/background.js', LF will be replaced by CRLF the next time Git touches it
warning: in the working copy of 'extension/budget.js', LF will be replaced
… [460 more characters trimmed]

Toda la extensión está ya en inglés y los presupuestos siguen en euros. Los cambios están guardados en git.

Semáforo de las tiendas: "You can afford it", "You can, but it's tight" y "Not now". El botón es I bought it, y los textos de "Thinking…" y de error también están en inglés.
Popup: el resumen del mes (Free money this month, Purchases logged, Left) y el enlace a Settings.
La frase de Gemma: ahora la escribe en inglés. Ejemplo de la prueba: "Okay, Pablo, that SonoMax Pro sounds… intense, but hey, €320 and €11 a day could work, just chill for a bit!"
Euros: EUR sigue siendo la moneda por defecto y las cantidades se escriben al estilo inglés, "€179.99".
Las instrucciones internas que le da la extensión a Gemma y los comentarios del código también están en inglés, porque el jurado puede leer el repo.
La tienda de prueba está en inglés con precios en euros. El README ya nombra el botón como I bought it en las dos versiones.

He vuelto a probar a Gemma con las instrucciones en inglés y sigue acertando en los tres casos:

Tienda demo: elige €179.99, no el precio tachado ni el envío.
Listado de búsqueda: dice correctamente que no es un producto.
Página en inglés: elige $749, no el precio de lista.

Sigue reconociendo tiendas en español, francés, alemán, etc., porque esas palabras de "añadir al carrito" se quedan a propósito.

La extensión sigue sin probarse en el navegador de principio a fin, porque eso tienes que hacerlo tú. Si ya la habías cargado, pulsa recargar en chrome://extensions. La tienda de prueba sigue en http://localhost:8000.

Lo que queda antes del cierre (5 oct, 06:59 UTC):
1. Que la pruebe tu amigo y apuntar su reacción.
2. Capturas o un GIF del semáforo en verde, amarillo y rojo.
3. Subir el repo a GitHub, cuando me confirmes que lo publique.
4. El post en DEV, que te dejo como borrador para que lo revises. ¿En inglés también?

You

ya estoy en chrome://extensions en modo desarrollador, ¿que hago ahora?

Agent

▸ Bash curl -s -o /dev/null -w "ollama %{http_code}\n" -H "Origin: chrome-extension:...

Input

$ curl -s -o /dev/null -w "ollama %{http_code}\n" -H "Origin: chrome-extension://x" localhost:11434/api/tags; curl -s -o /dev/null -w "demo %{http_code}\n" localhost:8000/

Output

ollama 200
demo 200

Ollama y la tienda de prueba están funcionando. Sigue estos pasos:

1. Cargar la extensión
- Arriba a la izquierda, pulsa "Cargar descomprimida" (en inglés, "Load unpacked").
- En la ventana que se abre, ve a esta carpeta y pulsa "Seleccionar carpeta":
~\Documents\Hacktoberfest\2026\DEV Challenges\1\extension
Elige la carpeta extension, no la 1 que está por encima.
- Aparecerá la tarjeta Should I Buy y se abrirá sola la página de configuración. Si no se abre, en la tarjeta pulsa Detalles y luego Opciones de extensión.

2. Configurarla
- Rellena los campos con los números de tu amigo, o con unos de prueba como Income 1500, Fixed expenses 900, Savings 100 y Payday 1.
- Abre "Local AI (Ollama)" y pulsa Test connection. Debe decir "All set ✔".
- Pulsa Save.

3. Fijarla en la barra
- Pulsa el icono de la pieza de puzle 🧩, arriba a la derecha de Chrome, y luego la chincheta 📌 junto a Should I Buy. Así verás el icono del semáforo para abrir el resumen del mes.

4. Probarla
- Abre http://localhost:8000. Abajo a la derecha saldrá "🤔 Thinking…" y, unos segundos después, el semáforo con los auriculares a €179.99.
- Prueba también en una tienda real, por ejemplo la ficha de cualquier producto de Amazon o PcComponentes.

Si en algún paso ves un error en rojo en la tarjeta de la extensión o el semáforo no aparece, pégame lo que te sale o mándame una captura.

You

veo bien la aplicación, tengo varias capturas, voy a crear en github el repo

Agent

¡Genial que funcione! Al crear el repo en GitHub:

Hazlo público, porque el reto pide que se pueda ver el código.
Déjalo vacío: no marques "Add a README", ni licencia, ni .gitignore. Ya los tenemos en local y, si GitHub crea los suyos, chocarían al subir.
Un nombre que encaja: should-i-buy.

Cuando lo tengas, pásame la URL y lo subo yo con tus 3 commits. Si prefieres hacerlo tú, son estos comandos desde la carpeta del proyecto:

git remote add origin https://github.com/TU_USUARIO/should-i-buy.git
git branch -M main
git push -u origin main

Y las capturas: si me dices en qué carpeta las tienes, las meto en el repo (en docs/screenshots/) y las enseño en el README antes de subirlo. Quedan muy bien para el jurado y luego las reutilizamos en el post de DEV.

You

https://github.com/Virien84/Should-I-Buy.git

Agent

▸ Bash cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && git ls-remote https:/...

Input

$ cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && git ls-remote https://github.com/Virien84/Should-I-Buy.git 2>&1 | head -5; echo "---"; git remote -v

Output

---

El repo existe y está vacío. Subo el código:

▸ Bash cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && git remote add origin...

Input

$ cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && git remote add origin https://github.com/Virien84/Should-I-Buy.git && git branch -M main && git push -u origin main 2>&1 | tail -4

Output

branch 'main' set up to track 'origin/main'.
To https://github.com/Virien84/Should-I-Buy.git
 * [new branch]      main -> main

▸ Bash cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && git ls-remote origin ...

Input

$ cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && git ls-remote origin && git log --oneline

Output

fcd27f48922d8d357488cf86e34f04971ff53318 HEAD
fcd27f48922d8d357488cf86e34f04971ff53318    refs/heads/main
fcd27f4 Translate the whole extension and demo shop to English; keep budgets in euros
c96a6a0 Rename to Should I Buy; English settings page and bilingual README
59421c6 FinDeMes: extensión que dice si te puedes permitir un producto, con Gemma en local

El código ya está en GitHub: https://github.com/Virien84/Should-I-Buy. Lo que hay en GitHub coincide con tu copia local: los 3 commits en la rama main, con la extensión, la tienda de prueba, el README bilingüe y la licencia MIT. El primer commit todavía se llama "FinDeMes" en su mensaje; no pasa nada, es solo el historial.

Siguientes pasos, para los que necesito dos datos tuyos:

Capturas: ¿en qué carpeta las tienes? Las meto en docs/screenshots/, las enseño en el README y lo subo. Así el repo se ve bien cuando lo abra el jurado.
El post en DEV: ¿lo escribo en inglés, como el resto? Y si tu amigo ya lo ha probado, cuéntame qué dijo, aunque sea en una frase. Esa reacción da puntos extra y es lo que más luce en el post.

You

Las capturas están en ~\Documents\Hacktoberfest\2026\DEV Challenges\1\screenshots. Los nombres deberían ser bastante explicativos. El post de DEV escríbelo en inglés, como el resto. Mi amigo lo ha probado y le ha gustado, primero, que no tarda tanto, ya que mientras el mira las características del producto, el proceso piensa y el tiempo es óptimo, y lo segundo, es que agradece mucho el comentario realista y la estimación.

Agent

▸ Bash cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && ls -la screenshots; g...

Input

$ cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && ls -la screenshots; git status --short

Output

total 1612
drwxr-xr-x 1 david 197609      0 Oct  4 23:08 .
drwxr-xr-x 1 david 197609      0 Oct  4 23:06 ..
-rw-r--r-- 1 david 197609  72989 Oct  4 23:06 Data introduction.jpg
-rw-r--r-- 1 david 197609 286352 Oct  4 23:08 green light after bought it.jpg
-rw-r--r-- 1 david 197609 276901 Oct  4 23:07 green light thinking.jpg
-rw-r--r-- 1 david 197609 294467 Oct  4 23:08 green light veredict.jpg
-rw-
… [221 more characters trimmed]

Antes de subirlas las miro, por si se ve algún dato personal (números reales, pestañas, cuentas):

▸ Read ~\Documents\Hacktoberfest\2026\DEV Challenges\1\screenshots\Data introduction...

Input

~\Documents\Hacktoberfest\2026\DEV Challenges\1\screenshots\Data introduction.jpg

Output

[image: screenshot]

▸ Read ~\Documents\Hacktoberfest\2026\DEV Challenges\1\screenshots\green light think...

Input

~\Documents\Hacktoberfest\2026\DEV Challenges\1\screenshots\green light thinking.jpg

Output

[image: screenshot]

▸ Read ~\Documents\Hacktoberfest\2026\DEV Challenges\1\screenshots\green light vered...

Input

~\Documents\Hacktoberfest\2026\DEV Challenges\1\screenshots\green light veredict.jpg

Output

[image: screenshot]

▸ Read ~\Documents\Hacktoberfest\2026\DEV Challenges\1\screenshots\green light after...

Input

~\Documents\Hacktoberfest\2026\DEV Challenges\1\screenshots\green light after bought it.jpg

Output

[image: screenshot]

▸ Read ~\Documents\Hacktoberfest\2026\DEV Challenges\1\screenshots\yellow light vere...

Input

~\Documents\Hacktoberfest\2026\DEV Challenges\1\screenshots\yellow light veredict.jpg

Output

[image: screenshot]

▸ Read ~\Documents\Hacktoberfest\2026\DEV Challenges\1\screenshots\red light thinkin...

Input

~\Documents\Hacktoberfest\2026\DEV Challenges\1\screenshots\red light thinking.jpg

Output

[image: screenshot]

▸ Read ~\Documents\Hacktoberfest\2026\DEV Challenges\1\screenshots\red light veredic...

Input

~\Documents\Hacktoberfest\2026\DEV Challenges\1\screenshots\red light veredict.jpg

Output

[image: screenshot]

▸ Bash cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1/screenshots" && mv "Data ...

Input

$ cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1/screenshots" && mv "Data introduction.jpg" settings.jpg && mv "green light thinking.jpg" green-thinking.jpg && mv "green light veredict.jpg" green-verdict.jpg && mv "green light after bought it.jpg" green-bought.jpg && mv "yellow light veredict.jpg" yellow-verdict.jpg && mv "red light thinking.jpg" red-thinking.jpg && mv "red light veredict.jpg" 
… [21 more characters trimmed]

Output

green-bought.jpg
green-thinking.jpg
green-verdict.jpg
red-thinking.jpg
red-verdict.jpg
settings.jpg
yellow-verdict.jpg

▸ Edit ~\Documents\Hacktoberfest\2026\DEV Challenges\1\README.md --- old I built it ...

Input

~\Documents\Hacktoberfest\2026\DEV Challenges\1\README.md
--- old
I built it for a friend who never makes it to the end of the month.

### Why open, local AI
+++ new
I built it for a friend who never makes it to the end of the month.

### Screenshots

| 🟢 You can afford it | 🟡 You can, but it's tight | 🔴 Not now |
|---|---|---|
| ![Green verdict on Amazon](screenshots/green-verdict.jpg) | ![Yellow
… [300 more characters trimmed]

Output

The file ~\Documents\Hacktoberfest\2026\DEV Challenges\1\README.md has been updated successfully.

▸ Edit ~\Documents\Hacktoberfest\2026\DEV Challenges\1\README.md --- old La hice par...

Input

~\Documents\Hacktoberfest\2026\DEV Challenges\1\README.md
--- old
La hice para un amigo que nunca llega a fin de mes.

+++ new
La hice para un amigo que nunca llega a fin de mes.

Las capturas están en la [sección en inglés](#screenshots).

Output

The file ~\Documents\Hacktoberfest\2026\DEV Challenges\1\README.md has been updated successfully.

▸ Bash cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && git add -A && git com...

Input

$ cd "~/Documents/Hacktoberfest/2026/DEV Challenges/1" && git add -A && git commit -q -m "Add screenshots to the README

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>" 2>/dev/null; git push -q 2>&1 | tail -2; git ls-remote origin main; git log --oneline -1; curl -s -o /dev/null -w "%{http_code}\n" https://raw.githubusercontent.com/Virien84/Should-I-Buy/main/screenshots/green-verdict.jpg

Output

warning: in the working copy of 'README.md', LF will be replaced by CRLF the next time Git touches it
de12cff8308ab58f7bab2b6fcd5d246ee25b33f2    refs/heads/main
de12cff Add screenshots to the README
200

Las capturas ya están en GitHub. Ahora el post para DEV:

▸ Skill {"skill": "devrelay-publishing", "args": "Draft DEV post for Hacktoberfest We...

Input

{
  "skill": "devrelay-publishing",
  "args": "Draft DEV post for Hacktoberfest Weekend Challenge: Build for a Friend (#hf26challenge), project Should I Buy, in English"
}

Output

Launching skill: devrelay-publishing

▸ ToolSearch select:mcp__devrelay-gateway__create_article,mcp__devrelay-gateway__get_my_ar...

Input

select:mcp__devrelay-gateway__create_article,mcp__devrelay-gateway__get_my_articles,mcp__devrelay-gateway__list_tags,mcp__devrelay-gateway__get_article_by_path,mcp__devrelay-gateway__get_article_content

Output

Busco la plantilla oficial de entrega del reto y si ya tienes algún borrador:

▸ mcp__devrelay-gateway__search_dev_to_semantic Hacktoberfest Weekend Challenge Build for a Friend announcement submission te...

Input

Hacktoberfest Weekend Challenge Build for a Friend announcement submission template hf26challenge

Output

{"Ok":[{"id":4793441,"title":"This is a submission for the [Hacktoberfest Weekend Challenge: Build for a Friend]","description":"This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend           What I...","tags":["devchallenge","weekendchallenge","hf26challenge"],"path":"/devapraveen/this-is-a-submission-for-the-hacktoberfest-weekend-challenge-build-for-a-friend-1751","ur
… [2022 more characters trimmed]

▸ mcp__devrelay-gateway__get_my_articles {"state": "unpublished"}

Input

{
  "state": "unpublished"
}

Output

{
  "Ok": [

  ]
}

▸ mcp__devrelay-gateway__search_dev_to_semantic browser extension local LLM Ollama privacy personal finance budgeting

Input

browser extension local LLM Ollama privacy personal finance budgeting

Output

{"Ok":[{"id":4564332,"title":"A small privacy decision that changed my web calculator","description":"I built Calculator Toolkit for a narrow moment: someone has a restaurant bill, wants to check the tip...","tags":["webdev","privacy","ux","discuss"],"path":"/laconicleon/a-small-privacy-decision-that-changed-my-web-calculator-5c99","url":"https://dev.to/laconicleon/a-small-privacy-decision-that-ch
… [2170 more characters trimmed]

▸ mcp__devrelay-gateway__get_article_content {"id": 4779462}

Input

{
  "id": 4779462
}

Output

{"Ok":"Hacktoberfest is here, and so is the first of our [**five Hacktoberfest DEV Challenges**](https://dev.to/devteam/hacktoberfest-2026-dev-challenges-five-challenges-one-prompt-a-new-theme-every-week-1e54): a **[Weekend Challenge](https://dev.to/challenges/hacktoberfest-weekend-2026-10-01)** sized to fit into your weekend.  We've set the timing so that wherever you are in the world, you'll hav
… [9362 more characters trimmed]

Ya tengo la plantilla oficial y no tienes ningún borrador previo. Redacto el post:

▸ mcp__devrelay-gateway__create_article {"title": "Should I Buy: a traffic light for my friend who never makes it to ...

Input

{"title": "Should I Buy: a traffic light for my friend who never makes it to the end of the month", "description": "A browser extension that tells you, on any shop page, whether you can afford what you're looking at, powered by Gemma running locally with Ollama.", "tags": ["devchallenge", "weekendchallenge", "hf26challenge", "gemma"], "main_image": "https://raw.githubusercontent.com/Virien84/Shoul
… [9462 more characters trimmed]

Output

{"Ok":{"id":4797614,"title":"Should I Buy: a traffic light for my friend who never makes it to the end of the month","description":"A browser extension that tells you, on any shop page, whether you can afford what you're looking at, powered by Gemma running locally with Ollama.","tag_list":["devchallenge","weekendchallenge","hf26challenge","gemma"],"user":{"name":"David Anton Alvarez","username":"
… [9825 more characters trimmed]

Tienes el post en tu DEV como borrador, sin publicar:
https://dev.to/virien84/should-i-buy-a-traffic-light-for-my-friend-who-never-makes-it-to-the-end-of-the-month-mn3-temp-slug-7062961/edit

Las capturas también están ya en el README del repo en GitHub.

Cómo está hecho el post:
- Sigue la plantilla oficial del reto: What I Built, Demo, Code, How I Built It, Why Does Open Innovation Matter? y Prize Categories.
- Lleva las etiquetas obligatorias devchallenge, weekendchallenge y hf26challenge, más gemma. Opta a Best Use of Gemma.
- Incluye tus 6 capturas, cada una con una explicación, y el repo de GitHub embebido.
- Hay una sección "What my friend said" con sus dos comentarios: que no se hace lento porque el modelo piensa mientras él lee las características del producto, y que valora el comentario realista y la estimación.
- La parte técnica explica que Gemma lee la tienda pero las cuentas no las hace la IA, la precarga del modelo para que vaya rápido y el error 403 de Ollama con las extensiones.
- Dice que lo construiste con Claude Code como agente. DEV lo marca como "some AI", que es lo honesto.

Revisa estas cosas antes de publicar:
1. El género de tu amigo: en el post me refiero a él como "he/him" porque dijiste "ese amigo". Si prefieres que no quede claro o es una amiga, lo cambio.
2. La captura de configuración muestra tu nombre y 1900/1100/400. Si son tus números reales y no quieres que se vean, la quito.
3. "I built it in an evening": cámbialo si no te parece exacto.

Plazo: se entrega hasta las 08:59 hora española de mañana (5 oct). Se publica desde DEV con el botón Publish, o puedes pedirme que lo publique yo.

Opcional: el reto anima a enseñar la sesión con el agente y a los jueces les gusta. Puedo guardar esta conversación con DevRelay y añadirla al post en la sección "My Agent Session". ¿Quieres?

You

sí, guarda la sesión y añádela al post

Prize Categories

Best Use of Gemma: Gemma 3 4B runs locally through Ollama and does both AI jobs: reading arbitrary shop pages into structured product data (with JSON-schema constrained output), and writing the short, honest verdict sentence.