Our open-source devtool is trending #1 on Product Hunt! Beacon is self-improving memory for coding agents. 🫶 I’d love your support: → Turn past debugging sessions into reusable skills → Share learnings across Claude Code, Cursor, Codex, and 20+ harnesses → Search and replay past sessions → Stop wasting... tokens rediscovering solutions → Build a feedback loop that makes your agents better over time → Free, open source, and local-first We built Beacon because we were tired of teaching our agents the same things over and over, only for them to start from zero every new session. Every agent session should make the next one smarter. ⭐️ Star us on GitHub:show more

2 weeks since launching Beacon and our $5.5M seed announcement 🚀 → 3K+ GitHub stars → 300% Discord growth → 85%+ Cloud user growth → 6,200+ actionable findings The response has been incredible. Thank you all! Support us on Product Hunt ❤️

can't believe how much time and patience Beacon will save 😎

Glad to hear that Paul! Would love your feedback on the project :)

Upvoted on PH! I hope Beacon wins #1 🏆

thank you Peggy! 🙏 to the product hunt algo gods :)

this solves a problem developers encounter every single day 🙌

Thanks Gina! If you get a spare sec, please upvote our launch on product hunt and I'd love your feedback on Beacon :)

Really cool. I jump between T3 Code, Cursor and Codex, so this is great. Also open source. 🔥

Amazing Catalin! Would love your feedback on Beacon, let me know what you think :)

Sure thing!

Finally agents can learn from experience.

It's indeed finally here! Would really appreciate your support with an upvote on our launch :)

Let’s go 🦾

Thank you!!!

Let's go!!! 🚀🚀🚀

300% Discord growth in two weeks says people really needed this

I think so too :) Please join our Beacon community and keep the product feedback coming!

Turning past debugging sessions into reusable skills is the memory feature I actually miss. I keep my own notes in a searchable file, because an agent that starts from zero every session costs me the same lesson twice.

Oh wow, definitely check out Beacon -- I'd love to get your feedback on the product and hope it becomes a superpower for you :)

stars growing quickly!

Just getting started :) Thanks Joyce!

Sharing this with all my builder friends ASAP

Thank you Shushant! Grateful for your support :)

congrats on #1 open source makes this even better

We love OSS! I'd love your feedback on Beacon if you check it out Karan :) Don't forget to upvote our launch on Product Hunt:

LFG team

thanks Toma! really appreciate your support :)

n u got the support n upvote

Thanks Aaliya! Really grateful for your support 🙏

Huge congrats on number one Beacon sounds incredibly useful

Thank you Nova Tech! We're just getting started with Beacon, so much more to build :) Would really appreciate an upvote on our launch if you get a spare sec:

does the memory sharing work smoothly if i switch between cursor and claude code

Great work, Justin! This is exactly what coding agents need! 🙌 Curious how much time and tokens we could save if agents actually remembered what worked in previous sessions.

Exactly Csaba! We actually surface time + token savings in our agent session behavioural findings on our Beacon Cloud product :) Check it out here:

congrats on the launch. the cross-harness handoff alone is worth it for me, i bounce between claude code and codex way too much

Amazing, so glad to hear this Rez! I'd love to get your feedback on Beacon as you use it:

Sure, count me in :)

@shakshukan moving fast guys! love it 🚀

@shakshukan Yessir! Thanks Nirmal :)

sharing learnings across claude code and codex is the real hook. do old skills get pruned when the codebase changes, or does the memory just keep growing?

Great question Loki! Beacon uses traces from your agent sessions as a feedback loop to identify when a skill stops being useful or becomes stale as the codebase evolves. The goal is to update or retire outdated learnings, so agents get better over time instead of accumulating bad context :) Get started here:

coool

Thanks Loki!!

3K stars in two weeks is crazy. congrats Justin and team

Thank you!!! If you get a spare sec would really appreciate an upvote on our product hunt launch:

Memory built from debugging sessions goes stale fast. A fix learned on last month's code keeps getting applied after a refactor, so I'd tie each skill to the files it came from and expire it when those files change.

Totally agree Vincenzo! Tying skills to the files and codebase state they came from is a great signal. Beacon also want to use traces as a feedback loop. This means if a skill stops working after a refactor, Beacon will catch that and update or retire it automatically!

Turning old debugging sessions into skills is the part that actually saves tokens. How does Beacon decide a learned skill has gone stale after the code changes?

self-improving memory is a bold claim for a json file

super cool! upvoted :)

Really appreciate it Kanav! If you try Beacon out let me know, I'd love to get your feedback on the product :)

every bug fix and workaround carrying over across coding agents instead of starting from zero every time you switch tools. love startups solving the memory layer problem 🧠✨

LFG!🔥🤝

Thanks Vlado! We're just getting started 🚀🚀🚀

W!!

🫶

Congrats on the launch, the minimal footprint and response time look super clean!

Hoping it keeps the non-debugging stuff too. The tenth time you tell it which button style the app uses hurts more than any bug

100% Danylo! It's not just about debugging! Beacon will remember your coding conventions, UI preferences, architecture decisions, and all the little corrections you keep making. Give it a whirl here:

Self-improving memory is exactly what agents need

Love to hear it! Would really appreciate an upvote on our product hunt launch if you get a spare sec :)

love that its open source and local-first

Thanks David! Definitely try it out and let me know what you think (I'd be grateful for a ⭐️) :

handoff resume is useful. who signs the memory?

Thanks Olivia! Beacon has a configurable human-in-the-loop step so you can look at any memory candidates that are generated and sign off before they're automatically used by your agents!

This is really cool! Does Beacon also work across different projects? Or is the memory tied to a specific repo?

Looking at that trending product! 🌟 What a feat with Beacon. Impressive! 🚀