Tell Claude Code something today and Codex still knows it tomorrow. Bourdon gives your AI agents one shared memory instead of one per tool - a directory of plain YAML on your own disk, in a published schema, served over MCP. 16 agent integrations (Claude Code, Codex, Cursor, Copilot, Cascade…), recognition-first runtime (answers in milliseconds, not after a 3-second search), peer federation you must explicitly opt into. pip install bourdon.

I run multiple AI coding agents  (Claude Code, Codex, Cursor)  and the same
thing kept breaking my flow: every agent treats memory as retrieval. Ask about
a project and it goes silent for seconds, searches, then reads you a paragraph
from its notes. A human with the same context says "Oh, Project Alpha, the
file converter" in 200ms and fills in the rest while talking.

Bourdon is my attempt to close that gap. It's not a better RAG. it's a
runtime timing layer (L0 hot cache → L6 federation) plus one shared memory
file every agent reads and writes. Your memory is a directory of plain YAML
on your own disk in a published schema.  grep it, git it, or walk away with
it. Nothing lives in someone else's database.

I run multiple AI coding agents  (Claude Code, Codex, Cursor)  and the same
thing kept breaking my flow: every agent treats memory as retrieval. Ask about
a project and it goes silent for seconds, searches, then reads you a paragraph
from its notes. A human with the same context says "Oh, Project Alpha, the
file converter" in 200ms and fills in the rest while talking.

Bourdon is my attempt to close that gap. It's not a better RAG. it's a
runtime timing layer (L0 hot cache → L6 federation) plus one shared memory
file every agent reads and writes. Your memory is a directory of plain YAML
on your own disk in a published schema.  grep it, git it, or walk away with
it. Nothing lives in someone else's database.

bourdon

One memory shared by every AI you use. And it stays yours

One memory shared by every AI you use. And it stays yours