







hey everyone,
i’ve been working on nodex, a local-first notes app for people who write a lot, forget where things are, and don’t want their notes locked inside another cloud account.
the basic idea is simple:
- write markdown notes
- search the whole vault instantly
- drop quick thoughts from anywhere into a daily inbox
- open a visual board when a note needs messy thinking
- ask a local ai model questions across your notes
- everything stays on your machine
the part i’m most excited about is the daily capture flow. if i have a thought while doing something else, i can send it straight into today’s inbox note instead of opening the app and breaking focus.
i also added an inkboard mode for rough visual thinking, so it’s not only linear notes. sometimes a thought needs a board, not another paragraph.
demo attached. would love honest feedback, especially from people who have tried notion/obsidian/logseq and still feel like their system gets messy.
currently, the lifetime access to nodex license is available for just $5 onetime (this is only for the first 25 users, few have already claimed it, so ending soon).
you can get it here: https://nodexnotes.online
thought of building a local first privacy focused notes taking app. because why not?
why to upload our notes to cloud? why cant our notes be just with us?
so i built nodex.
it runs in the terminal. notes are written like markdown. search is local. if you already run local models, you can ask questions over your notes with llama.cpp.
that’s it.
current build has:
- markdown-style editing
- sqlite full-text search
- wiki links and backlinks
- local ask-vault using gguf models
- markdown import/export
- no cloud workspace
- no desktop app telemetry
basically, a simple and classy local first notes taking app for everyone.
i am giving out lifetime access to the first 25 users just for $5 (ending soon), this will include all the future updates.
the app will be actively maintained and updated by me based on user feedback and users needs.
you can watch the demo video and get the app here: https://nodexnotes.online
drop your honest feedback below.
hey, i’ve been building nodex, a local-first notes app that runs in the terminal.
the basic idea is simple:
- write markdown notes
- search your whole vault locally
- ask questions over your notes using local gguf models
- keep everything on your own machine
i just added a new feature called context radar.
while you’re writing, it checks the current note against the rest of your vault and shows:
- notes that look like duplicates
- notes that might be a better place for the current idea
- related notes you may want to link
- an option to create a merge draft instead of manually copy-pasting things around
it doesn’t delete or rewrite anything automatically. it just suggests what might be connected, and you decide what to do.
demo attached above.
would be interested to hear if this is actually useful for people who have lots of scattered notes.
Hey everyone, I’ve been building Nodex, a local-first notes app for people who want Markdown notes + search + AI without putting their private notes into a cloud app.
The idea is simple:
- Write notes in a terminal-style workspace
- Store everything locally
- Search the full vault with SQLite full-text search
- Ask questions over your notes using local GGUF models through llama.cpp
- No account for notes, no cloud sync, no telemetry
I built it because I wanted something closer to an IDE for thinking: fast, keyboard-first, Markdown-native, and private by default.
It is paid early access right now: $5 one-time. The app is private-source for now, but the notes and AI flow run locally on your machine. AI features require llama.cpp + a GGUF model; normal note-taking/search works without that.
Demo video is attached above.
If you want to get it, here's the website: https://nodexnotes.online
I’m looking for blunt feedback:
- Is terminal-first notes too niche? BTW YOU DONT NEED TO RUN COMMANDS IN TERMINAL TO OPEN THE APP, JUST DOUBLE CLICK ON THE APP TO OPEN IT, BUT IT RUNS IN THE TERMINAL.
- Would non-devs use this if setup was smoother?
- What would make local AI notes actually useful for you?
fine-tuned LiquidAI’s LFM2.5-230M on Fable-5 traces and shipped it as GGUF
tiny 230M coding-agent model. trained at 4096 ctx. exported Q4_K_M / Q8_0 / F16. runs locally.
repo:
fine-tuned LiquidAI’s LFM2.5-230M on Fable-5 traces and shipped it as GGUF
tiny 230M coding-agent model. trained at 4096 ctx. exported Q4_K_M / Q8_0 / F16. runs locally.
repo:
what should i do to improve my looks and appearance more?
rate and advice
i got a random idea of why should we sacrifice our privacy by keeping notes in apps like notion, etc.
so thought of building a simple local first rust based vim like notes taking app with local AI models integrated in it for RAG.
is it a good idea?
the demo video is uploaded above, check it out and lmk.
should i release it for one time payment of $5? or just open source it?
open for all kinds of feedback.
need some advice to improve my looks
(even better if from a female, bros are welcomed too)
give me some specific advice to looksmax
need looksmaxxing advice from others perspectives
as i cant know/think from others perspectives and know what they think about me lol
i have built and used a lot of agent harnesses.
i found out one thing:
- the harnesses which depend the “LEAST” on the LLMs often give the best performance
also the harnesses that almost always depend on the LLMs are wrappers and not harnesses.
your harness needs to use LLM for decision making, and very complex reasoning stuff, not all the trivial stuff.
thats what separates wrappers from good harnesses.
what do you think?
rust is basically following “prevention is better than cure”. complexity now = lesser complexity at runtime or production.
what do you think?
i started building an inference engine in rust, and realized how hard it is to write good rust code without AI assistance, but the complexity makes it even better, complexity now = lesser complexity at runtime or at production. basically, rust follows prevention is better than cure.
btw i have finished implementing a gguf file parser + a tensor engine (all the matrix ops) for my inference engine in rust. will soon update once i finish building it.
btw i am a rust beginner, started learning it like 2 weeks ago.
open to your advices, suggestions to learn it faster and build more cool stuff. kudos!