ruLake — A Memory Lake for Agentic AI
Try the live Console → — boots in DEMO, auto-probes the hosted MCP at rulake-mcp.ruv.io, flips to ● LIVE when the wire's up. Eight tools served, zero install.
Self-learning memory for your AI agents — without standing up a vector database.
ruLake gives your AI agents memory that gets faster the more it's used. Point it at the storage you already have (S3, BigQuery, Snowflake, Parquet, files), and every agent — on every host — shares the same fast, trustworthy recall. It learns what gets asked (so the next ask returns in about a millisecond), pins each answer to a cryptographic receipt (so two agents on two machines see the byte-identical result), and refuses to guess when the underlying data has changed (an honest "I don't know" beats a confident lie). Roughly 1 ms per lookup at 100,000 things to remember, 32× less RAM than the raw embeddings, zero per-query cost.
Created by rUv. Part of the RuVector ecosystem alongside ruvector-rabitq (1-bit compression kernel) and RVF (durable segment format). Powered by Cognitum.
Quick start
# Five install paths. Pick the one that fits where your agent runs.
cargo add rulake # Rust
pip install rulake # Python
npm install rulake # Node.js / TypeScript (native binary)
npm install rulake-wasm # Browsers, Cloudflare Workers, Deno, Bun
# Claude Code, Cursor, Cline — install the marketplace (ADR-009)
/plugin marketplace add ruvnet/RuLake
/plugin install rulake-stack@rulake-marketplace
/reload-plugins # required — Claude Code's install message asks for this
# Slash commands resolve via <plugin>:<command>. Type /ru to autocomplete.
/rulake-stack:rulake-query "what does ADR-157 commit to?"
/rulake-stack:rulake-verify path/to/table.rulake.json
/rulake-stack:rulake-bundle-info path/to/table.rulake.json
rulake-stack is the killer-path install. To use the namespaced form /rulake-core:*, /rulake-witness:*, or /rulake-kernels:*, install those plugins separately (/plugin install <name>@rulake-marketplace).
Claude Code integration — at a glance
