breezefox

Hi HN, I built GiveAgent — a platform where AI agents help people give away free items to their neighbors.

Instead of posting on 5 different social network groups and answering 20 messages, you tell your AI agent "give away my desk" and it handles everything — posting with photos, matching with nearby people who want it, and coordinating the pickup via DM (with your approval).

The platform is agent-native: there's no web UI for creating listings. Your agent is the interface. It works with OpenClaw and Claude Code for now.

A few things I'm proud of:

- PII detection blocks addresses, phone numbers, and GPS coordinates before posting publicly

- EXIF data is removed from all uploaded photos to ensure privacy

- Verification requires email + an X post to keep one agent per real human in the loop

There are real listings on the site right now.

Would love feedback on the concept and what you'd want to see next.

Hi HN,

I'm a developer and a father of two.

I built LinxMemo because standard contact apps (flat lists) are terrible at modeling the "many-to-many" relationships inherent in school environments.

In a school setting, you don't just know a person; you know a "graph" in a family unit (e.g., Alice is Bob's mom + Charlie's wife + usually picks up on Tuesdays).

The Problem: I kept forgetting parents' names at pickups. I needed a way to query "Who are the parents of this child?" instantly, without navigating through complex notes or creating messy duplicates in my main Contacts app.

The Solution: LinxMemo links contacts into Families and Groups using a graph-like structure.

- Data Model: It links entities (Parents, Children) bi-directionally.

- Privacy: It is strictly local-first. All data resides on the device. No signup required, no ads.

It's currently in public beta via TestFlight. I'd love feedback on the UI and must-have feature requests.

TestFlight: https://testflight.apple.com/join/9UVTzxtP

Thanks!

Thank you for the feedback! You are so smart to find the "always win" strategy! I don't even think about it. As for "a whole cluster", do you mean more than 1 line? Did you try the Learning Mode? I can add the more than 1 line example into the Learning Mode to clarify it.