Beacon turns agent sessions across Claude Code, Cursor, Codex, OpenCode, and 20+ other harnesses into a compounding knowledge layer, where every successful run
makes every future agent smarter.
brew trust asymptote-labs/tap && brew tap asymptote-labs/tap && brew install beacon && beacon endpoint installFrom the community
Another insane Jev use case! Jev is making it dramatically cheaper to evaluate what actually happened inside an agent run. And finally, someone open-sourced a self-improving memory layer that can put that signal to work across agent harnesses: - Claude Code - Codex - Cursor - Show more
x.com/i/article/2101…
Your JEV agents can finally learn from each other Beacon uses Jev to find the valuable lessons buried in agent sessions and turn them into reusable skills across Claude Code, Codex, Cursor, and 20+ other harnesses
JEV IS INSANE your coding agents forget everything the second the session ends. Every fix. Every dead end you already hit. Every "no, not like that." Gone. Tomorrow you teach it all over again. Someone just open sourced the fix, and it runs on Jev. It's called Beacon, by Show more
x.com/i/article/2101…
Another insane Jev use case! Jev makes it incredibly cheap to evaluate and classify agent runs at scale. And finally, someone open-sourced a self-improving memory layer that can put that capability to work across agent harnesses. It turns your agent sessions into a compounding Show more
x.com/i/article/2100…
最近JEV太热了，我觉着这个项目是JEV最重要的应用场景。终于有人干实事了！@asymptotelabs 开源了 Beacon，一个自我进化的 agent 记忆层。 它把你散在 Cursor、Claude Code、Codex、OpenCode 等 20 多个工具里的会话全打通，靠 Jev 便宜地筛出真正值钱的运行，变成能复用的技能。agent Show more
这50个神奇的网站，知道的人都在偷偷玩👇 跟你说句掏心窝子的话，网上真正上头的好东西，早就不在那几个天天给你灌信息流的App里了。真正的宝藏，全躲在下面这50个「兔子洞」里，一脚踩进去就别想爬出来。先收藏，早晚有你用得上的那天。 先说地球这颗球，实时到吓人： 1️⃣
A friend sent me this thing this morning too
Another insane Jev use case! Jev is making it dramatically cheaper to evaluate what actually happened inside an agent run. And finally, someone open-sourced a self-improving memory layer that can put that signal to work across agent harnesses: - Claude Code - Codex - Cursor -
Great idea by continuously captureing your agent history across harnesses and uses Jev to identify which runs are actually worth learning from. It then turns the highest-signal workflows, corrections, and debugging patterns into reusable skills. Actually we are using jev the Show more
Another insane Jev use case! Jev is making it dramatically cheaper to evaluate what actually happened inside an agent run. And finally, someone open-sourced a self-improving memory layer that can put that signal to work across agent harnesses: - Claude Code - Codex - Cursor -
hell yes, we're so stoked on this use case
Ever wondered if you have to switch between claude-code, codex, hermes & openclaw to work on something and you wish it would all be at one place? Beacon is an open-source memory layer for AI coding agents. It captures session history across → Claude Code, → Cursor, Codex, Show more
Another insane Jev use case! Jev is making it dramatically cheaper to evaluate what actually happened inside an agent run. And finally, someone open-sourced a self-improving memory layer that can put that signal to work across agent harnesses: - Claude Code - Codex - Cursor -
memory needs to follow me, not separate tools. Great to see efforts in this direction
Another insane Jev use case! Jev is making it dramatically cheaper to evaluate what actually happened inside an agent run. And finally, someone open-sourced a self-improving memory layer that can put that signal to work across agent harnesses: - Claude Code - Codex - Cursor -
Beacon logs the full run across Claude Code / Codex / Cursor; Jev scores which corrections are reusable skills. One edge-case fix in Claude can stop Codex from relearning it. That's the cross-harness memory layer I actually wanted.
Another insane Jev use case! Jev is making it dramatically cheaper to evaluate what actually happened inside an agent run. And finally, someone open-sourced a self-improving memory layer that can put that signal to work across agent harnesses: - Claude Code - Codex - Cursor -
turning high signal corrections into reusable skills across 20+ harnesses!
Another insane Jev use case! Jev is making it dramatically cheaper to evaluate what actually happened inside an agent run. And finally, someone open-sourced a self-improving memory layer that can put that signal to work across agent harnesses: - Claude Code - Codex - Cursor -
Most agent memory dies with the session. @_avichawla flags Beacon by @asymptotelabs: open-source history across @claudeai Claude Code, @cursor_ai, @OpenAI Codex, @opencode, and 20+ harnesses. Workflows and fixes stay reusable.
Another insane Jev use case! Jev is making it dramatically cheaper to evaluate what actually happened inside an agent run. And finally, someone open-sourced a self-improving memory layer that can put that signal to work across agent harnesses: - Claude Code - Codex - Cursor -
The folks at @asymptotelabs are onto something, Beacon is very cool
Another insane Jev use case! Jev is making it dramatically cheaper to evaluate what actually happened inside an agent run. And finally, someone open-sourced a self-improving memory layer that can put that signal to work across agent harnesses: - Claude Code - Codex - Cursor -
Another insane Jev use case! Jev is making it dramatically cheaper to evaluate what actually happened inside an agent run. And now, someone has open-sourced a self-improving memory layer that turns those insights into reusable knowledge across agent harnesses: > Claude Code > Show more
x.com/i/article/2101…
a correction you make in Claude Code shouldn’t need explaining again in Codex. Beacon by @asymptotelabs is an open-source memory layer across coding agents. it captures sessions and lets you use Jev to flag reusable corrections and debugging patterns. you review and approve the Show more
Esto es el missing layer: memoria que sobrevive al cambio de harness. Beacon + Jev — sesiones messy in, skills out. Bookmark si ya saltas entre Cursor / Claude Code / Codex. github.com/Asymptote-Labs…
Another insane Jev use case! Jev makes it incredibly cheap to evaluate and classify agent runs at scale. And finally, someone open-sourced a self-improving memory layer that can put that capability to work across agent harnesses. It turns your agent sessions into a compounding
Insane to see how harnesses are jumping guns blazing with variety of niche use cases, personally this has been my favourite so far as it uses Jev as the right control layer to emulate and learn from the best agent trace in a compounding memory landscape
Another insane Jev use case! Jev makes it incredibly cheap to evaluate and classify agent runs at scale. And finally, someone open-sourced a self-improving memory layer that can put that capability to work across agent harnesses. It turns your agent sessions into a compounding
Most agent memory tools record everything and call it learning. This one uses Jev to decide what's worth keeping. That's the whole difference.
Another insane Jev use case! Jev is making it dramatically cheaper to evaluate what actually happened inside an agent run. And finally, someone open-sourced a self-improving memory layer that can put that signal to work across agent harnesses: - Claude Code - Codex - Cursor -
Beacon learns from your work across AI coding agents, identifies the workflows and corrections worth keeping, and makes that knowledge reusable by future agents.
Your agent knowledge compounds across tools instead of disappearing when a session ends.
Claude Code, Cursor, Codex, OpenCode, Cline, and 20+ more harnesses in one continuous history.
Keep useful workflows, corrections, debugging patterns, and repository conventions after a session ends.
Review knowledge once, then let future agents retrieve it through MCP or Agent Skills.
See prompts, responses, tools, commands, edits, approvals, MCP activity, and tokens in one trace.
Own durable JSONL data, choose explicit destinations, and move between harnesses without lock-in.
Deploy your way
| Capability | Local | RecommendedBeacon Managed |
|---|---|---|
| Cross-harness agent telemetry | ✓ | ✓ |
| Local, CI & cloud collection | ✓ | ✓ |
| Unified agent session history | ✓ | ✓ |
| Store data locally | ✓ | ✓ |
| Forward to your own data stack | ✓ | ✓ |
| Hosted ingest & storage | — | ✓ |
| Search & analytics | — | ✓ |
| Identity mapping | — | ✓ |
| Agent session retention | You manage | Unlimited |
| SSO & RBAC | — | ✓ |
Compounding agent memory
Beacon captures what happens across your AI agents and turns successful workflows, corrections, and debugging patterns into reusable knowledge that compounds across harnesses.
Agent observability
Beacon gives you a complete view of agent activity across harnesses, repositories, models, and workflows so you can see what agents are doing, where they succeed, and where they get stuck.
Workflow debugging
Beacon traces every command, tool call, retry, and failure across a session so you can quickly pinpoint what went wrong and improve the workflow behind it.
