It's a small, silly protest at the stupidity of the Online Safety Act that just came into force. The ID's actually work (for Reddit, Discord etc.) which highlights how terrible this implementation is.
So I built Piper: basically AI that makes phone calls for you. You tell it what you need (book appointment, check on an order, dispute some charge, whatever), and it handles the entire conversation while you do actual work. Right now it's a web app, Chrome extension is pending approval but soon you'll be able to click any phone number anywhere and just let Piper handle it.
Technical stuff that was harder than expected:
Latency - every millisecond counts in conversation, had to optimize around kv cache, got it down to ~1000ms to first word over PSTN for telephony, which feels pretty natural
Keeping the voice agents on track - built custom context engineering logic that constantly updates the agent's situational awareness, so it knows when it's been transferred, when it's on hold, etc
Done ~50 successful calls with early testers so far. Main failures are when they need complex verification or documents. Also had to take down our IVR navigation temporarily :/, found some edge cases that were causing unnecessary transfers but working on fixing that.
I really think we're heading toward this world where AI talks to AI for most routine things, and phone calls might be the first real example of this happening at scale!
you can check out the a voice demo on our website. https://pipervoice.com
I wanted to make some photomosaics for an anniversary gift, but I ended up building this tool and turning it into a free website that anyone can use.
For those who don’t know, a photomosaic is an image made up of many smaller tile images, arranged in a way that forms a larger, recognisable picture.
The best part? Everything runs directly in your browser. No files are uploaded, and there’s no sign-up required.
Kiln is a free, open tool that gives you everything most AI projects need in one integrated package:
- Eval system: including LLM-as-judge evals, eval data generation, human baselines
- Fine-tuning: proxy to many fine-tuning providers like Fireworks/Together/OpenAI/Unsloth
- Synthetic data generation: deeply integrated into evals and fine-tuning
- Model routing: 12 providers including Ollama, OpenRouter, and more
- Git-based collaboration: projects are designed to be synced through your own git server
The key insight is that these tools work much better when they're integrated. For example, the synthetic data generator knows whether you're creating data for evals vs. fine-tuning (which have very different data needs), and evals can automatically test different prompt/model/fine-tune combinations.
It runs entirely locally - your project data stays in local files, and you control your own git repos. No external services required (though it integrates with them if you want).
Main project GitHub: https://github.com/Kiln-AI/Kiln
Demo GitHub where I use it to build a 'natural language to ffmpeg command' demo with evals, fine-tunes, and synthetic data (including demo video): https://github.com/Kiln-AI/demos/blob/main/end_to_end_projec...
I’ve been collecting and analyzing expired domains for years — especially those about to drop. Every day, tens of thousands expire. Most are junk, but a few still have traffic, backlinks, SEO value, or just great names. Finding them used to take hours.
Last week I put my internal tools online: https://pendingdelete.domains No login, no paywall Updated daily Combines domain history, traffic, SEO data and AI-driven insights to identify valuable expirations The goal: help spot valuable domains quickly and skip the noise.
Still a work-in-progress — would love feedback: Is this useful? What signals or filters would you add? Any UI or speed improvements?
Thanks!
Our goal is to make institutional-grade research accessible to everyone. We'd love feedback and features you'd would like to see.
So over a weekend, I built h‑codex – a tool that can pull context from scattered repos into Cursor / Claude Code (integrated via MCP); ensuring that they’ve got the full picture when doing the plans/implementation.
How it works:
- Indexes the repos - Chunks code with AST for optimal chunk boundaries - Generates embeddings and stores them in pgvector for fast semantic search
Check it out: https://github.com/hpbyte/h-codex
I’ve been working on a side project called SocialKit, a simple API for scraping structured data from public social media posts (YouTube, Shorts, more coming soon).
It's designed for dev, no-code users, and marketers who want to extract things like summaries, transcripts, details, and more from YouTube.
I sold two previous side projects (LectureKit and CaptureKit), and this is my latest attempt at building something useful and focused. I would love to get any feedback :)
Happy to answer any questions, and thanks for checking it out!
Jonathan
GitHub issues for popular projects are often flooded with "+1”, "Me too”, and other unhelpful comments, making real discussions hard to track. This is a simple Google Apps Script (taking reference from https://github.com/cline/cline/discussions/581#discussioncom...), which scans your Gmail inbox for emails triggered by these "noisy” comments and automatically trashes them.
Also accompanying (as comment below the script) is browser console script to dim or hide these comments from issue discussions.
---
See the spam condition on this thread: https://github.com/cline/cline/discussions/581
I built AllEars, an Android app that listens for real-world sounds — like snoring, coughing, barking, alarms, glass breaking, and more — and lets you automate actions on your phone using simple IF–THEN flows.
It runs fully on-device, with no cloud, no mic streaming, and no internet requirement. Your phone detects sound events locally using TensorFlow Lite (YAMNet) and executes your defined actions.
Examples:
IF snoring AND coughing THEN vibrate + log timestamp
IF glass break THEN trigger webhook
IF dog bark OR doorbell THEN send notification
Each flow uses flat AND/OR logic, supports multiple actions, and can include time filters. The app runs as a low-power background service with a persistent notification for transparency.
Available actions:
- Play a sound
- Vibrate
- Send a notification
- Log a timestamp
- Trigger a webhook
Why I built it: I wanted user-programmable automation driven by ambient sound — without sending any audio to the cloud.
Think of it as IFTTT for offline sound events.
Try it on Play Store: https://play.google.com/store/apps/details?id=com.cliqueraft... — no signup needed. Free plan includes 3 flows.
Looking for feedback on:
1. What sounds would you want to detect and respond to?
2. What other actions/triggers would make this genuinely useful (keeping things offline and safe)?
3. Any UI feedback on the flow builder logic?
Thanks! Happy to dive into anything technical
Fun fact: Font Awesome icons are viewed 6–8 TRILLION times per month, so we have a pretty good idea which ones are used the most! Armed with this knowledge, we added seven entirely new icon packs, each with 200+ of the most frequently used icons in their own unique designs.
What's new in v7:
- 300+ new icons
- Redesigned core icons for better visual consistency, clearer shapes, and smoother outlines
- The Icon Wizard now supports over 40 modifiers and can instantly convert modified icons into Duotone
- Better rendering and file loading
- Seven new Pro+ icon packs: Chisel, Etch, Jelly, Notdog, Slab, Thumbprint, Whiteboard, and more on the way
Head to https://fontawesome.com/ to browse all of the new icons and styles.
We love feedback and I'm happy to answer any questions about the release, the tech behind it, or icon design in general! (Note that I primarily work on a sister product called Web Awesome, so it might take me a little time to reply if I have to ping a colleague about a specific question!)
Code here: https://github.com/tekaratzas/GuitarGPT
Uses Gemini 2.5 Flash to analyze video of you playing, Highlights segments then pumps out a personalized practice routine.
Been using it to get unstuck on some more difficult passages. Pretty insane how good it is given this is just the start.
After building numerous FastAPI backends, I consistently found myself repeating the same configuration, including auth, email, payments, migrations, etc.
In order to handle all the boilerplate, I developed FastLaunchAPI, a production-ready startup package.
Included are:
JWT authentication with social login and email
Webhooks and Stripe billing together
Alembic, PostgreSQL, and SQLAlchemy
Background work for celebrities
SMTP email that includes templates
Integration of LangChain and OpenAI
Setup for Pytest + API documents
There is a comprehensive documentation, which can be viewed before pruchase.
I would be happy to hear any comments or inquiries!
Appcircle CodePush is a fully managed, modern alternative built into Appcircle’s end-to-end mobile CI/CD platform.
Why Appcircle CodePush?
- No app store resubmissions: Instantly deliver updates to users while keeping your security and compliance standards.
- Enterprise-ready:
Self-hosted or cloud deployment options
Role-based access control, SSO/LDAP support
Code signing, package diff, code obfuscation support, and end-to-end encryption
Detailed audit logs and compliance reporting
- Flexible usage:
Manage OTA updates through the Appcircle dashboard, with the CLI, or by adding Appcircle CodePush as a workflow step in your CI pipeline, depending on what fits your process best.
- End-to-end management:
Centralized dashboard for updates, rollbacks, and channel management
Automated workflows, built-in testing, update history
Fast and scalable delivery with secure CDN support
- Transparent and trackable:
Monitor all updates, rollbacks, and usage from a single dashboard
Access compliance-ready reports and full audit trails
- Get started:
Docs & quickstart: https://docs.appcircle.io/code-push
Happy to answer any questions or feedback!
This is still experimental but provides a way to add webhook support to MCP. You can clone the repo and try the streamable_http_webhook examples.
It’s designed for AI systems, developers, and ops teams working in complex, multi-database environments.
Highlights: - Unified schema model across database types - Built-in schema version control - Zero-downtime migration and replication - Quantum-safe encryption and data obfuscation - AI agent integration via Model Context Protocol (MCP)
Our demo runs:
Gemini 2.5 Flash as the fast, lightweight model
Qwen3-Coder-480B as the more powerful planner and code generator
But you can configure whatever combo works best for your use case. Forge offers better performance, lower costs, and full control with self-hosting options.
You can find our quickstart repo here: https://github.com/TensorBlock/claude-code-forge
Would love to hear your thoughts on this!
I built Side Reader because I got tired of the constant tab-switching between PDF readers and AI chat - copying text, switching tabs, pasting, repeat. It was driving me crazy.
Over time, I added more features like custom highlight buttons, screenshot functionality, and local storage.
I recently launched a prototype of *ClearDoc*, an AI-powered tool to extract structured data from unstructured documents like invoices, bills of lading, certificates, etc.
It uses *OCR (PaddleOCR)* and *LLMs* to detect and align key fields — even for complex documents with tables, nested fields, or in different languages.
It doesn't require templates and can be *self-hosted* (demo runs on my own GPU).
Live demo (no sign-up): http://cleardoc.v5ent.com/ Demo video: https://www.youtube.com/watch?v=u83T6iewfNs
Right now: - Fields are auto-aligned visually on the document - Works with PDFs, images, scans - No custom field design/editing in the demo yet
Would love feedback on: - Which use cases matter most to you? - What would make this valuable enough to adopt?
Thanks!
After chatting with my cofounder, we were both intrigued. So I opened up Cursor and hacked together a prototype. After a few iterations, we had a working alpha: by dropping an OpenAPI file, we were able to get an AI Agent that understands it and can call those APIs using natural language. The Agent understood every prompt, made the right API calls and even auto-corrected bad payloads.
For internal APIs, auth is always tricky, so we added a system that supports API keys and credentials client-side only. Tokens are never stored on our backend. We also integrated it with Postman, so you can easily import your collections.
We imagine different use-cases and plan to add integrations (Stripe, Slack, CRMs, etc) and allow users to embed it where they want (for example on their platforms, so that their end-users can chat with the APIs and perform actions).
Looking for early users and feedbacks.
What it does: - Upload your resume.
- Set your minimum salary requirement (optional).
- Get an email with personalized job matches that actually pay well.
- Includes AI-researched salary data (base + equity) and company ratings
- LLM for resume parsing
- LLM for embeddings/matching
- PostgreSQL + pgvector for similarity search
- Dynamic job fetching via Google.
Currently works for all english speaking countries (not just US) and provides salaries in local currency.
I haven't figured out monetization yet - currently burning through my own money on API costs. Would love ideas on this. Some thoughts: - Charge job seekers after X matches?
- Recruiting/sourcing tool for companies?
- Premium features (more frequent updates, salary negotiation tips)?
- The onboarding experience
- Email format/content
- What other features would be valuable?