HN user

siegers

310 karma
Posts57
Comments39
View on HN
github.com 1mo ago

Strands Shell: Give your agent a shell without giving it keys to your machine

siegers
9pts0
kiro.dev 1mo ago

New in Kiro Web: Build with Spec, GitLab, and More

siegers
5pts1
github.com 1mo ago

Agent Toolkit for AWS

siegers
5pts0
www.anthropic.com 1mo ago

How we contain Claude across products

siegers
6pts0
kiro.dev 2mo ago

Introducing Kiro Web: Build, delegate, and steer right from the browser

siegers
4pts0
strandsagents.com 2mo ago

What We Learned from One Year of Building Production Agents

siegers
2pts0
hyperframes.heygen.com 2mo ago

HyperFrames lets AI agents compose videos by writing HTML, CSS and JavaScript

siegers
5pts0
github.com 2mo ago

Supabase Feature Preview: RLS Tester

siegers
2pts0
chromewebstore.google.com 2mo ago

Show HN: Readox, playlist-style read aloud extension for articles, PDFs and docs

siegers
4pts2
readox.ai 3mo ago

Show HN: Readox – Turn web pages and PDFs into a playable reading queue

siegers
4pts0
jvogel.me 3mo ago

One Method Was Using 71% of CPU. Here's the Flame Graph

siegers
11pts2
www.tama96.com 3mo ago

Show HN: Tama96 – A virtual pet for your desktop, terminal, or AI agent

siegers
5pts0
www.jmduke.com 3mo ago

A month of OpEx quick wins

siegers
3pts0
www.xiegerts.com 3mo ago

Somewhere Between Prompt and Context Engineering

siegers
2pts0
jvogel.me 4mo ago

Java is fast, code might not be

siegers
228pts249
www.agentpng.dev 4mo ago

Show HN: Agentpng – turn agent sessions into shareable images

siegers
4pts1
www.xiegerts.com 4mo ago

Hit Your 1 Rep Max with AI

siegers
3pts0
kiro.dev 5mo ago

Kiro: DeepSeek, MiniMax, and Qwen now available as open weight model options

siegers
6pts1
www.addtosheets.com 6mo ago

Show HN: Workflow automation builder for saving website data

siegers
4pts0
deepmind.google 7mo ago

Gemma Scope 2: open suite of tools for language model interpretability

siegers
2pts0
kiro.dev 8mo ago

Kiro is generally available: Build with your team in the IDE and terminal

siegers
7pts0
www.addtosheets.com 8mo ago

Show HN: Add to Sheets Chrome Extension for Google Sheets – No Code Automations

siegers
4pts0
kiro.dev 1y ago

Kiro.dev - The AI IDE for prototype to production

siegers
85pts6
github.com 1y ago

Opencode – AI coding agent, built for the terminal

siegers
6pts1
www.xiegerts.com 1y ago

Bring Your Own API Key: User-Provided LLM Keys and Prompts in Chrome Extensions

siegers
2pts0
www.addtosheets.com 1y ago

Show HN: Use Chrome Keyboard Shortcuts to Save Selected Text to Google Sheets

siegers
1pts1
chromewebstore.google.com 1y ago

Show HN: A Chrome extension to save to Google Sheets in one click

siegers
4pts0
www.lemonsqueezy.com 1y ago

Lemon Squeezy – Payout fees sliced

siegers
1pts0
gluestack.io 1y ago

Gluestack-UI v2: React and React Native Components and Patterns (NativeWind)

siegers
1pts0
chromewebstore.google.com 1y ago

Show HN: You're online all day. Why not use the browser to remind you of things?

siegers
2pts1

Hi! Yes, the premium voices are Kokoro. I’m only exposing the English voices right now because the rest of the pipeline around them is English-first and custom, especially pronunciation/G2P, QA, and timestamp awareness. I’d like to expand that over time, but I don’t want to overpromise multilingual support before the surrounding stack is ready. So I'm taking it one language at a time based on demand and feedback.

AI summaries are currently generated remote, not local. Those currently leverage gpt-4o-mini. TTS and OCR are on-device and summarization is the cloud-backed feature.

I like this approach and that it's so flexible and approachable. My nit for the site would be to explain it a bit better - it's a bit hard to grok with everything going on until I went to the repo. There's also opportunity for non-tech users here just using the agent skill but I'm not sure they'll understand the use if they read through the site.

I had made a skill for something similar to use just HTML5, Canvas and JavaScript, and export the video to MP4 (https://github.com/siegerts/skill-canvas-video).

Claude Opus 4.7 3 months ago

Agree. I keep effort max on Claude and xhigh on GPT for all tasks and keep tasks as scoped units of work instead of boil the ocean type prompts. It is hard to measure but ultimately the tasks are getting completed and I'm validating so I consider it "working as expected".

Claude Opus 4.7 3 months ago

I enjoy switching back and forth and having multi-agent reviews. I'm enjoying Codex also but having options is the real win.

(dev here) Thanks for the feedback. We choose circle to remain consistent with other todo applications (Todoist, iOS Reminders, etc). Semantically, these aren't selection boxes or choices per se. They are to indicate a completion action outside of the scope of a form element.

(dev here) Thanks for the feedback! Most of this is related to sustainability of the platform as a whole. When we originally posted on HN, that was the number one theme - future proofing the platform for users (this also included the subscription aspect) since the majority of the app hadn't gone through any major updates since the initial launch (~3.5 yrs ago). This was just to remain transparent in where we have been focusing our energy.

Thanks! Yeah, I wanted to keep it specific for GitHub but can see how the base functionality can be used for something more generic. On each page match hit for a signed in user, the API is checked to see if the user has notes - if the matching were expanded beyond GitHub, then that would need to be taken into account to reduce ‘ping’ requests that check if notes exist.

(dev here) Completely understand the concern. This uses an embedded Google File picker to allow you to select what Sheets to allow access to. Until you do that, the extension doesn't have r/w access. The integration with Google (OAuth) has been verified and gone through their review process to only request/use minimal perms for the extension to function as expected.

https://www.addtosheets.com/blog/how-to-save-data-to-google-...

(dev here)

Something I've done different for this extension that I haven't done in the past:

- had an email waitlist that got some interest for months leading up to the actual release

- engaged with potential users in community channels (reddit/discord(s)/etc) to get early feedback and keep them in the loop

- have buttons in the profile dropdown menu within the extension for "Provide feedback" and "Leave a review"

All of these have made a big different, in general, to the overall process. I started doing this months before the extension was published.

Happy to answer any questions.

(cofounder)

Thanks and glad to hear! Our main goal was to _not_ try to make it "everything" and while still making it enjoyable to use. We'd appreciate any feedback once you do some testing.

Send over to lowercasehq@gmail.com or ping us on X @lowercase_app.

Thanks! So we actually started with offline first originally (a standalone desktop app) and then rebuilt for the web. We had plans to revisit based on interest, etc. so it's good to get the feedback here and use that to shape the direction.

(cofounder here)

Thanks for the feedback! We have some ideas, and to be honest, the feedback here has been great to firm up some of our assumptions.

I'd like to follow up on this if you're interested. Feel free to email (lowercasehq@gmail.com).