Launching today
beacon.sh
Open-source self-improving memory for AI coding agents
577 followers
Open-source self-improving memory for AI coding agents
577 followers
Beacon turns agent sessions across Claude Code, Cursor, Codex, OpenCode, and 20+ other harnesses into a compounding knowledge layer, so every agent starts smarter than the last. It captures every agent trace across your environment and turns activity into clear operational visibility. Get started for free at beacon.sh
Launch Team / Built With
Framer AI AgentsDesign and publish professional sites with AI
Promoted
Does Beacon keep the memory local, or sync it to a cloud account?
@modiyilreji Hey Reji! We have 2 modes: local-only (no data leaves your machine), and Beacon Cloud (your agent session data is synced to your cloud account. With Beacon Cloud, the main value add we're giving users is 1) continuously generating findings + skill opportunities for your sessions, and 2) no infra overhead for you to think about. With Beacon Cloud you can also choose different redaction modes (metadata-only or full agent traces).
I'd love your feedback on Beacon :) Definitely check out our repo and if you like it please give us a star! https://github.com/Asymptote-Labs/agent-beacon
This is like AI agents having Apple Continuity. Very cool.
Que: how do you handle the artifacts across agents?
@5harath We normalize all AI telemetry (and extend OpenTelemetry's genAI standard) into a standard data model across 20+ harnesses! This is what enables us to show you all your AI activity in one place + make your context + memories portable across harnesses! More on our coverage here: https://github.com/Asymptote-Labs/agent-beacon/tree/main#supported-agents
Does it ever save the wrong context and bring it back later when you don’t actually need it?
Congrats @justindsouza & team!
Thanks for sharing @fmerian :)
@fmerian @hamza_afzal_butt really great question Hamza! This is candidly one of the hardest problems we've seen as we've built out Beacon. We have a configurable human-in-the-loop step so that you can approve any reusable memory + skill candidates that are generated. In addition to that, we leverage your agent sessions as a feedback loop to actually evaluate whether those approved memories + skills are used at runtime.
Would love your feedback on Beacon if you check it out! https://github.com/Asymptote-Labs/agent-beacon
What happens to a saved workaround when a repo change makes it obsolete? That seems like the tricky part of carrying lessons between sessions.
@oleksii_sekundant Great question Oleksii! This is definitely one of the hardest problems we're tackling. Beacon uses traces from subsequent sessions as a feedback loop to see whether a saved workaround still works. As the repo evolves, we can detect when that knowledge becomes stale and update or retire it.
I'd love your feedback on our approach. Please check out our open-source project and let me know what you think (don't forget to star)! github.com/Asymptote-Labs/agent-beacon
The portability angle is the part that hits home for me — I switch between Claude Code and Cursor in the same day, and right now the second tool starts from zero no matter what the first one figured out. Turning past sessions into reusable skills instead of another summary doc is the right shape for that. One thing I'm curious about: are those memories scoped per repo by default? A workaround that makes sense in one codebase could be actively wrong in another, so I'd want the compounding layer to know which project a lesson came from.
@jcbuilds The scoping is really up to you and dependent on your codebase layout :) You can choose where to persist the memories + skills that Beacon generates for you, and can either share across repos or scope per repo.
Definitely try it out and let me know what you think JC! I'd love to get your feedback on the product: https://github.com/Asymptote-Labs/agent-beacon
I love the LLM concept and dev. However the intro video although very cinematic provides no information about what you have built. Best of wishes on your launch. Cheers!!
@richatsealedvault fair point haha! ICYMI @justindsouza recorded a live demo here. hope it clarifies
@justindsouza @fmerian Yes, this should be the first video on your page.
@justindsouza @richatsealedvault thanks for your support and feedback, it means the world 🙏
one more thing: please help us spread the word on X, repost this
@richatsealedvault Thanks Richard for the kind words! Definitely check out our demo video, or even better our open-source project and let me know what you think :) I'd love your feedback: https://github.com/Asymptote-Labs/agent-beacon
Congrats on the launch @justindsouza! Finally I can switch between Claude Code and Codex without re-explaining the same context all over again 🙌
@peggy_rayzis thank you Peggy! And yes, exactly -- that superpower is finally here :)
@justindsouza @peggy_rayzis it's about time haha