Ежедневные Show HN

Upvote0

Show HN за 29 ноября 2025 г.

18 постов
176

Nano PDF – A CLI Tool to Edit PDFs with Gemini's Nano Banana #

github.com favicongithub.com
40 комментариев8:44 PMПосмотреть на HN
The new Gemini 3 Pro Image model (aka Nano Banana) is incredible at generating slides, so I thought it would be fun to build a CLI tool that lets you edit PDF presentations using plain English. The tool converts the page you want to edit into an image, sends it to the model API together with your prompt to generate an edited image, then converts the updated image back and stitches into the original document.

Examples:

- `nano-pdf edit deck.pdf 5 "Update the revenue chart to show Q3 at $2.5M"`

- `nano-pdf add deck.pdf 15 "Create an executive summary slide with 5 bullet points"`

Features:

- Edit multiple pages in parallel

- Add entirely new slides that match your deck's style

- Google Search enabled by default so the model can look up current data

- Preserves text layer for copy/paste and search

It can work with any kind of PDF but I expect it would be most useful for a quick edit to a deck or something similar.

GitHub: https://github.com/gavrielc/Nano-PDF

31

oeis-tui – A TUI to search OEIS integer sequences in the terminal #

github.com favicongithub.com
2 комментариев1:20 PMПосмотреть на HN
i always loved looking up the On-Line Encyclopedia of Integer Sequences (OEIS) when researching a sequence of numbers.

so I decided to make a TUI and CLI for it so that I can browse sequences in the terminal.

it supports almost all the features on the site (including the OEIS Webcam) and supports graphs, a preview pane, exporting and bookmarks.

more features here: https://github.com/hako/oeis-tui?tab=readme-ov-file#features

repo: https://github.com/hako/oeis-tui

gitHub releases: https://github.com/hako/oeis-tui/releases

cargo: cargo install oeis-tui

18

Zero-power photonic language model–code #

zenodo.org faviconzenodo.org
7 комментариев6:45 PMПосмотреть на HN
The model uses a 1024-dimensional complex Hilbert space with 32 layers of programmable Mach–Zehnder meshes (Reck architecture) and derives token probabilities directly via the Born rule.

Despite using only unitary operations and no attention mechanism, a 1024×32 model achieves coherent TinyStories generation after < 1.8 hours of training on a single consumer GPU.

This is Part 1 - the next step is physical implementation with $50 of optics from AliExpress.

16

No Environment Setups Anymore #

gitarsenal.dev favicongitarsenal.dev
17 комментариев7:57 PMПосмотреть на HN
Hi everyone, for last 7 months, I have been learning all the attempts made to eliminate codebase environment setups.

Here's my product which is a leap in the same direction and will help you run any codebase on relevant machine.

Check it out on gitarsenal.dev/ and we got ranked 6th on Product Hunt as well.

4

TimeTracker – Self-hosted time tracking with invoicing (120 features) #

timetracker.drytrix.com favicontimetracker.drytrix.com
1 комментариев8:53 AMПосмотреть на HN
I've been building TimeTracker over the past few years, and it's grown into a comprehensive self-hosted platform for time tracking, project management, and invoicing.*What it does:*- Time tracking with persistent server-side timers- Professional PDF invoice generation- Project & task management- Client CRM- Expense tracking- Advanced reporting & analytics- Role-based permissions- REST API*Why self-hosted:*I was tired of paying $15-20/month for time tracking tools and having my data locked in SaaS platforms. This gives complete control.*Tech:*- Python/Flask backend- PostgreSQL + Redis- Docker deployment- 120+ features documented*Deployment:*One docker-compose command. Works on any server, including Raspberry Pi.Would love feedback from the HN community! What features would be most valuable for you?
3

I built 19 AI agents because one wasn't enough to coach my workouts #

arvo.guru faviconarvo.guru
2 комментариев9:43 AMПосмотреть на HN
Hey HN,

  I've been lifting for years and got frustrated with fitness apps that either (a) generate a static plan and forget about you, or (b) require manual logging with no intelligence.
 
  So I built Arvo—an AI personal trainer that adapts after every single set, not just at the start of your workout.
 
  Why 19 agents instead of one? Each handles a specific job:
  - Exercise Architect (selects movements based on equipment/goals)
  - Load Navigator (calculates weight adjustments based on RIR)
  - Volume Manager (tracks MEV/MAV/MRV per muscle group)
  - Pattern Scout (extracts insights from natural language notes)
  - Movement Adapter (biomechanical weight conversion when swapping exercises)
 
  They coordinate to make decisions in <500ms.
 
  Example: You finish set 1 of bench press, felt easy (RIR 4). Arvo bumps weight +2.5kg for set 2. Set 2 was a grind (RIR 1). Arvo keeps the weight but suggests longer rest.
2

ClearHearAI-The Essential App for Hearing Impaired and Deaf Communities #

clearhearai.com faviconclearhearai.com
0 комментариев10:31 PMПосмотреть на HN
I built ClearHearAI to help hearing impaired and deaf people. It is a transcription app that provides context indicators (e.g. when questions are asked, urgent keywords are detected, conversation is happening). All audio processing happens entirely on your device - your voice never leaves your computer. Conversation transcripts are stored locally on the device.

Any feedback welcomed.

2

Lightweight macOS menu bar Pomodoro Timer #

github.com favicongithub.com
0 комментариев1:01 PMПосмотреть на HN
I was using an app called Notch Flow, but even though I was a licensed user and paying every month, it kept asking me to re-enter my license from time to time.

So I used Claude Code and built a similar Pomodoro app in about an hour. My goal was to have something super minimal that could run for long hours without putting any load on the Mac.

I’m leaving the GitHub link below. I’d really appreciate it if you try it out and share your feedback.

https://github.com/berkaycit/pomodo-timer