Beacon is a local-first memory layer that captures coding-agent sessions across Claude Code, Cursor and Codex and serves reusable knowledge over MCP or Skills.
Yes, Beacon is open source under the MIT license.
Beacon is free to use.
Yes, Beacon can be self-hosted (the source is available under the MIT license).
Beacon is an MIT-licensed, local-first memory layer that captures AI-coding-agent sessions across harnesses — Claude Code, Cursor, Codex, OpenCode and others — and serves reusable knowledge over MCP or Agent Skills. It has 1,493 stars, is actively releasing (v1.3.x this week), and self-hosts. Its claim to grade runs into reusable skills is the tool's own, not a verified result.
Teams rarely standardize on one coding harness — some people use Claude Code, some Cursor, some Codex — and the useful context each agent builds up stays trapped in that tool. Beacon's angle is a shared, owned memory layer: capture sessions across harnesses, keep them local-first, and serve the accumulated knowledge back to any of them over MCP or Agent Skills. For a team building agent-driven internal tooling, that cross-harness, owned memory is the difference between per-tool amnesia and a shared knowledge base. The honest read: 1,493 stars and active releases make this the most-validated tool in this batch, but the specific claim that it grades runs into reusable skills is the tool's own and unproven — and it surfaced via a viral tweet whose hype hook the repo does not back up. Adopt it for the shared-memory mechanics; verify the "reusable skills" quality yourself.