TECH Signal 243 2 feeds carried it
OKF Agent Memory stores AI agent knowledge as Git-native Markdown with sub-300µs search
OKF Agent Memory provides a Git-native, vendor-neutral persistent memory layer for AI agents using plain Markdown files with YAML frontmatter, achieving sub-millisecond search and zero API costs.
AI coding agents lose context when sessions end, and existing solutions rely on vector databases or unstructured files. OKF Agent Memory offers a standardized, version-controlled alternative that lives in the repository, making agent memory auditable and free to retrieve. Engineers can adopt it without external services, reducing cost and complexity in agent toolchains.
Written by elseif from the cluster below · every claim links back to a sourceThe three things worth knowing
OKF Agent Memory stores knowledge as Markdown with YAML frontmatter in a knowledge/ directory, using the OKF v0.2 standard.
The Go-based tool provides in-memory BM25 search with sub-300µs latency and ~4ms graph validation, avoiding vector embedding API costs.
It enforces search-before-write and progressive disclosure to reduce context bloat and memory rot, with a CLI and MCP server for integration.
THE CLUSTER