Asymptote-Labs/agent-beacon

The cross-harness, self-improving memory layer for AI agents. Join our community: https://discord.com/invite/zdNChS2fBu

What it solves

Beacon solves the problem of "amnesia" in AI coding agents. Currently, when you use different coding tools like Claude Code, Cursor, or OpenCode, each session is isolated. Lessons learned, debugging patterns, and repository-specific conventions disappear once a session ends. Beacon ensures that knowledge compounds by capturing and sharing these insights across all your coding tools.

How it works

Beacon acts as a universal memory layer that sits beneath your various coding agents. It continuously records your session history—including prompts, tool calls, file edits, and approvals—and normalizes this data into a single, structured format. This history is stored locally as JSONL files. You can then use "Agent Skills" or MCP (Model Context Protocol) to allow future agents to search and retrieve this past knowledge, effectively giving your agents a persistent, shared memory.

Who it’s for

It is designed for developers who use multiple AI coding agents and want to maintain a consistent, evolving knowledge base across their entire workflow. It is ideal for those who want to avoid repeating the same corrections or debugging steps across different tools.

Highlights

  • Cross-harness support for 20+ coding tools including Claude Code, Cursor, Cline, and Codex.
  • Local-first architecture that keeps your data portable and under your control.
  • Unified event model that captures everything from token usage to specific file changes.
  • Extensible "Lenses" that allow you to visualize and analyze your session history.
  • Flexible output options to forward telemetry to SIEM or log aggregation platforms like Datadog, Splunk, or AWS S3.

Related

  • Project
    Dicklesworthstone/cass_memory_system

    cass‑memory is a CLI that aggregates session logs from many AI coding agents, extracts structured rules (playbook) with confidence tracking, and provides JSON‑based queries so any agent can retrieve relevant knowledge before coding.

  • Project
    memorax-ai/memorax-code

    A shared memory layer for AI coding agents that preserves architectural context, engineering lessons, and user preferences across sessions.

  • Dispatch
    Beacon: Bringing Visibility to Local AI Agents

    An exploration of Beacon, an open-source layer for local AI agent visibility, addressing the critical need for observability in the proliferating landscape of autonomous agents.

  • Project
    vshulcz/deja-vu

    deja-vu is a local, private memory engine for coding AI agents. It indexes transcripts from 41+ agent harnesses (Claude Code, Codex, Cursor, etc.), stripping secrets, and provides instant search, recall, and MCP integration so agents remember past fixes, commands, and file changes across sessions and tools. It features a fast CLI/TUI, proactive session-start recall, and benchmarks showing 97.2% R@5 on LongMemEval-S.

  • Project
    iAmCorey/Wake

    Wake is a native Rust‑based desktop app that aggregates, indexes, and lets you search the local conversation histories of many AI coding agents (Claude Code, Codex, Cursor, etc.). It offers a unified UI, full‑text search, one‑click resume of sessions, remote syncing over SSH, and a read‑only MCP/CLI for agents to query the history.