2026年6月13日 的 Show HN

I'm the developer of Trace, a non-intrusive, shortcut-driven Mac app that records and transcribes your meetings on-device. I know, another meeting transcription app. Please bear with me though, I'm confident that this is at least a little novel.

I primarily built Trace for myself. I'd been using MacWhisper, but there was enough fiddling before each call that I'd forget to start it and walk out of an hour-long meeting with nothing written down. So the things I cared about most were that it's quick to activate and stays out of the way. You activate Trace by pressing a global shortcut (configurable), which reveals a small bar at the bottom of your screen (there's also a keystroke and/or option to hide it entirely if you'd rather not see it at all).

As I was building it I wanted to bake in a couple of workflows I'd wished for in other transcription apps.

1. Mid-meeting you can press another global shortcut to mark a "key moment" and type a note. The note shows up in the resulting transcript inline at that timestamp. I wanted to add this because I kept catching myself thinking "wait, that bit matters" in meetings and reaching to jot it down in a separate app like Obsidian, which I then needed to add context to, which took me out of the meeting. I use it all the time. If I paste the transcript into an LLM afterwards (which I find myself doing more and more these days) the important moments are flagged so it doesn't gloss over them. This is more noticeable in longer meetings with lots of topics. 2. With another keyboard shortcut you can summon a rough live recap (subtitles, basically) to quickly recap what's just been said.

Trace uses standard macOS microphone and system recording APIs to capture both sides of the conversation as two separate tracks and then runs the system side through on-device diarization to identify speakers. Right now we only label them as "Speaker 1", "Speaker 2", etc but there are plans for speaker labelling in the future. You can also show a "live recap" as the call is happening to review what someone just said.

All transcription models run on your machine. To be clear though, Trace doesn't do any of the summarising itself, it just produces a markdown transcript, so if you want summaries then you need to pass the output to an AI.

The app is sandboxed and your audio/transcripts are never uploaded anywhere - they just exist as audio files and markdown on disk. The only network call Trace is required to make is on the first run to download the speech and speaker models (around 500MB) from Hugging Face, and after that it can be used fully offline. If enabled, a Google Calendar integration can auto-name sessions but that needs a network connection.

The app is £9.99 on the macOS App Store. I've been using it every day for months now and I'm super happy with how it's improved my workflow. Feedback very welcome.

I built Paca out of pure passion—a free and lightweight Jira alternative written in Go where humans and AI agents work together as equal teammates to plan sprints and assign tasks to each other. It is fully customizable with custom views, fields, and a WASM-based plugin architecture. My team uses it daily for our own development, so it will be continuously maintained and completely free forever

Dear Hacker News,

I'm kindly asking for your participation in the open beta for my AI-managed mini-games website. Thank you in advance!

For a limited time window, I'm setting the all-free feature flag to true. I hope you have a lot of fun exploring the AI's sense for games! Here and there, I tweaked it to help with visual consistency.

I would be deeply grateful if you opted into analytics.

$2,300 in API tokens...

Cheers!

Attached is my attempt at making a small toy S3 client without any other dependency besides libcurl and OpenSSL. Was tested mainly on MinIO (RIP) locally, so I would expect some bugs when using it against AWS, although I was able to play with it on some open access buckets

Be aware that I am not a C++ programmer and this project was indeed done to learn a bit of C++ myself :')

Feedback on any of the code, either on gtest, or the benchmarking section or the core itself is welcome!

I've been using the popular React Markdown renderers with Katex and Prism.js for rendering my Markdown and LaTeX, but was tired of having to bundle 300kb+ of min+gzip JS only for this (1.2MB+ of plain JS!). So I created a small Markdown renderer that does it all in a tiny package.

I added a small playground to the homepage, please feel free to try it and let me know what you think!

It's not perfect, it's definitely not "correct" in that I'm using Regex internally instead of a proper AST parser, but for my usecase and the majority of Markdown out there, this works perfectly fine (cue the StackOverflow post [1]). It's also conservative for this reason; no HTML by default, parsing wrong content produces escaped HTML entities instead of XSS.

[1] https://stackoverflow.com/a/1732454

feed-repeat is a small CLI/service that repeats entries from RSS/Atom feeds into new feeds. It fetches posts from a source feed, selects a random subset preferring older posts, and inserts them in an output feed.

I made feed-repeat for myself. I wanted a tool to send me old blogposts written by my favourite authors so that I could enjoy them again, but also not be overwhelmed by a wall of hundreds of unread old posts. feed-repeat does just that.

It can be self-hosted easily using the provided NixOS module or systemd service. It can also be made to run entirely serverless on GitHub Actions/Pages using the provided workflow. The website contains detailed docs and guides.

Built a free, private web UI for Ollama. You can chat with cloud models using your own API key, or connect it to a local Ollama instance running on your machine. Everything runs in the browser — streaming replies, charts, in-browser Python/JS execution, and GitHub repo context. No account required, nothing stored on our servers.

1. Headless mode

Headless mode allows you to use the AI as a command-line utility for automation and scripting. In Claude Code you run it with the -p flag: claude -p, in codex - exec, opencode - run.

2. Ask human

The traditional communication channel with the operator won't work in headless mode - we need to implement a dedicated tool. Here is an example of how this can be done https://github.com/sermakarevich/claude/tree/main/mcp/ask_hu...

3. Tasks queue

Beads is a lightweight distributed graph issue tracker for AI agents, powered by Dolt. You can create tasks, define dependencies between tasks, and have status, priorities, hierarchy. Beads helps prevent multiple tasks from being claimed by > 1 worker.

4. Worker artifacts

We want to be able to monitor how a worker is doing, at what stage it is, and resume it after a restart. For every task we can create a dedicated folder using the beads task id and put into it what we need. I put there: - plan and status md - knowledge md - events.jsonl - stderr

The worker is instructed in its prompt to check if artifacts exist, which allows it to proceed from where the job was left.

5. Worker isolation

To prepare to run multiple workers we need to isolate them. Git worktree can be used here. I am testing this approach: - worker gets the task and implements it - the next worker, spawned automatically, validates the task is done, tests it, merges the worktree, closes the ticket and creates another one for a fix if required

6. Multiple workers

To be able to run multiple workers we need a simple orchestrator. An infinite loop constantly checking beads / config and triggering new workers when required.

7. Coder agnostic

A worker can be basically any coder. I started with Claude, added Codex and Agy. And last added Opencode.

8. Subscription limits.

3 coding agents can burn the Claude $200 subscription limit in 30 minutes even if you switch to Sonnet 4.6. API tokens cost x40 compared to tokens in the subscription - this is too expensive. The idea I am testing is: - use the strongest model possible to analyse/design and add tasks - use a local model as a worker - use a stronger model to validate workers and add new tasks to fix potential misimplementations

I am using the qwen3.6:36B local model with Ollama, deployed on 2 GPU cards, 36GB in total, with a 256K context window. This is slower, but it is free of charge. And surprisingly it worked, and worked way better than I would expect it to. Fable 5 was extremely great at creating clear and simple tickets until it was.

Another approach I was considering is Bedrock qwen, paying per token, or renting a 96GB GPU for $1400 per month.

I found that it's optimal to run 3 workers concurrently even though Ollama processes 1 request at a time. The reason is the ask_human tool. If a worker asks me something at night - it has to wait until morning doing nothing. Running three +/- guarantees GPU load at 100%.

9. Nice integrations

UI - to observe tasks / beads / config / chat / analytics

It's easy to miss when a model asks a question. It's visible in the UI - a green circle near chat, but that's it. So I added a Telegram integration - now I receive questions from workers on Telegram and can reply there, get the status of tasks, create new tasks etc.

I am doing this for my PoC projects ofc: - improving fleet - building a data collection and analysis related app

What I am seeing is that 24x7 coders are closer than I thought they are. Even weaker models can deliver good results when the task is simple and well defined. All components for building these systems are there.

Repo: https://github.com/sermakarevich/fleet

A few weeks ago I posted an early version of this. It was just an MCP doing web search for AI assistants. Since then it evolved rapidly.

The problem I kept running into is that when I use an LLM for research, it gives me citations that look right but aren't most of the time. Sometimes the paper doesn't exist. Sometimes it does exist but it's been retracted. Most times the sources are not fully vetted nor truly relevant to my search. I got tired of checking these manually.

That's why I built this updated MCP. It looks the work up in Crossref to confirm it's real. It checks the Retraction Watch data to see if the paper was pulled. It also does a basic match between the claim I'm making and what the source actually says, so I know which references to read before I trust them.

You can also hand it a whole bibliography in BibTeX, RIS or CSL-JSON and it tells you which entries don't resolve or have been retracted. Dead links get saved to the Wayback Machine so the reference still points somewhere later.

It's a single Go binary. Open Source (MIT) and runs against most respected search engines including specialized domain-specific ones.

If you have the same problem, I hope this is useful to you too. Appreciate your feedback.

Was in search for a premiere pro alternative for linux, so I created this in 4-5 feedback and bugfix iterations with the help of fable 5. It's written in C++, uses Qt6 and ffmpeg and has most features that I needed for doing some basic video editing on linux. It's by far not perfect and only supports a small subsets of the features that software like premiere offers, but for me its good enough; and actually in some cases (surprisingly) even a little more performant then the original software for some playback tasks. It's still quite early but i would be interested in some feedback on this piece of software. Let me know what you think!