Why only gemini embeddings?

gemini's free tier is pretty generous, and no credit card required to use the free api. however OpenAI embeddings support is on the roadmap if u checked the repo!

interesting but why not just use memory.md / CLAUDE.md?

Those are per-tool and static (only for claude). they are dumped into context every session whether relevant or not, and they are not structured. DecisionNode is tool-agnostic and retrieval is explicit: the AI searches when it needs context, not on every session. Also works across Claude Code, Cursor, Windsurf simultaneously from one store. i still use both claude.md AND decisionnode during development, each one serves a different purpose in my workflow, they complement each other pretty well

How does this handle context that becomes stale? E.g., if one tool updates a deployment state and another tool's memory still has the old version. MCP gives you the pipe but not the consistency model.