Plain-Markdown long-term memory shared by Claude Code and Codex
Emerged finds new products worth a look, daily — what’s new today
A single-binary memory database for AI agents with a two-verb query language
Agents get persistent, ranked memory without running a vector or graph database, at recall close to mem0 and Graphiti on LoCoMo.
Fraise stores what agents learn as a temporal graph of facts, topics and entities, queried with just remember and recall. Recall is ranked and capped, favouring recent facts, because the caller pays for every token read back. Hybrid text, vector and graph retrieval runs inside one MIT-licensed binary, installable via Homebrew, Docker or packages, with an MCP server for Claude Code and Codex. The author reports 0.893 recall@10 on LoCoMo at 0.18s p50 against mem0, Letta, Graphiti and cognee.
Plain-Markdown long-term memory shared by Claude Code and Codex
An open plain-text file format for AI memory that any model can read
Shared memory for Claude Code and Codex sessions, with an LLM judge at the door
Shared, permissioned memory for coding agents, with an audit row per read
A single-file graph store you can query as it stood on any past date
Open streaming memory for real-time voice agents, split into facts and feelings