Nimbalyst Memory

Why Nimbalyst Memory in Nimbalyst?

Projects accumulate knowledge faster than any one conversation can hold: design docs, plans, decisions, tracker history, agent instructions. Nimbalyst Memory indexes that markdown into a local shadow index and exposes it to AI agents over MCP, so an agent can recall the relevant design decision or plan section instead of rediscovering it from scratch.

Everything stays on your machine. The index is a derived artifact — rebuildable at any time from your files — and agents can also record durable facts that persist across sessions.

Key Features

• Project knowledge index — Design docs, plans, CLAUDE.md files, and tracker content become searchable memory.
• Hybrid retrieval — Fast keyword and semantic search tuned for agent recall, served over MCP.
• Durable facts — Agents can remember and recall discrete facts that outlive individual sessions.
• Local and rebuildable — The index lives on your machine and can be rebuilt from source files at any time.

Use Cases

• Grounded agent sessions — Agents answer “how does X work here?” from your actual design docs.
• Long-running projects — Decisions made months ago stay findable without re-reading everything.
• Cross-session continuity — Facts recorded in one session inform the next one.