agent-memory
Plain-Markdown long-term memory shared by Claude Code and Codex
Emerged finds new products worth a look, daily — what’s new today
Strata
Shared memory for Claude Code and Codex sessions, with an LLM judge at the door
Teams running several coding agents get one memory they all read, where a judge admits or declines each write and keeps the reason, so bad notes stop spreading.
Strata lets multiple coding agents, across Claude Code and Codex, read and write memory scoped to a project area. Each contribution passes an LLM judge that records why it was admitted or declined, and a local console shows what the fleet believes and where each belief came from. It is a local Python service on SQLite and markdown, installed with pipx. MIT licensed, at v1.13.0 after 633 commits; the maker publishes adversarial test results and two known gaps.
More like this
Willder
Shared, permissioned memory for coding agents, with an audit row per read
Thrum
A local message bus that lets your coding agents review each other's work
Fraise
A single-binary memory database for AI agents with a two-verb query language
Openmsg
A shared inbox so coding agents from different vendors can message each other
Imza
A tiny native terminal agent where you co-write the spec before any code