💰 ViBo — Memory for AI Agents + Living Document Archive

One license. One key. Your agent remembers — and your documents think.

📖 The Living Archive: How My Agent Started Working Through ViBo — read the story

What it is

ViBo Memory — persistent memory for AI agents: facts survive between sessions, found by meaning, encrypted L1/L2/L3. Agent reads only relevant facts → 50-150× fewer tokens on large bases (measured).

ViBo Archive — a living document archive in our own .vibo format: compressed (~84%), searched by meaning, answers questions in milliseconds. LLM reads 22 KB instead of 146 KB.

How to use

vibo add "client Anna" "prefers espresso, budget 10K"
vibo find "what does Anna prefer?"

vibo archive pack ./contracts -o archive.vibo
vibo archive search archive.vibo "total amount and deadline?"
vibo archive unpack archive.vibo -o ./restored

Skill — drop into any Python agent (Hermes, OpenClaw, LangChain).
n8n — verified node n8n-nodes-vibo.
vibo-selfdeed — free autonomous mission skill: the agent restores context, finds and fixes problems safely, iterates via paths A/B/C and saves lessons back into ViBo memory. Learn more → · npm install vibo-selfdeed · ClawHub
Dify — plugin (PR in review).
Cloud API — memory for your SaaS.

Install requirements

Python 3.11 exactly. The core ships as compiled modules for CPython 3.11; 3.10 / 3.12+ refuse to import.
Install dependencies from requirements.txt in this repo, not by hand. usearch is pinned to 2.15.0 — newer builds crash the compiled index on the second add.
Check the host before trusting it: python3.11 tools/preflight.py <package-dir> tells you what actually works (memory core, index, add-ons).
Ubuntu 22.04 / Debian 12 (glibc < 2.38): memory works, but the Living Archive, license check and memory-tools modules do not load. Matrix and workarounds: docs/INSTALL_UBUNTU_22_04.md.
CI: .github/workflows/smoke.yml installs the released package, runs add → find → reindex and fails if a fact is not found — the regression that made 2.1.2 look empty on a clean machine.

What it is for

Banks — "What did we promise this client?" → milliseconds. Secrets stay encrypted (L3).
Archives — millions of pages stop being dust.
Libraries — "What is this book about?" → instant.
Law firms — a question instead of hours of searching.
Medicine — patient histories, protocols, studies.
Business — knowledge bases that answer 24/7.

Honest measurements

Memory	Tokens saved
100 facts	~2×
1K facts	10-20×
10K+ facts	50-150×
100K+ facts	up to 2,000×

Web search: 96-99%. Threads: -72%. Archive: 84%, 0.1s. Empty memory = 0 savings.

Pricing

One license, $5/month (first 100 users keep $2.5 forever). Cloud API $5/10/25. 2-day free trial.

Get it from @ViBomemorybot or wwwvibo.com.

Dust is no longer dust. Dust is memory that was waiting for its hour.

🚀 Projects

Project	What it is	Link
ViBo Memory	Persistent memory for AI agents — 50-150× fewer tokens, L1/L2/L3 encryption	site
ViBo Archive	Living document archive (.vibo) — 84% compression, search by meaning, answers in ms	demo
Cloud API	Memory for your SaaS — instant, encrypted, zero infrastructure	site
n8n node	Verified node for n8n workflows	npm
Dify plugin	Memory tools for Dify (search, add, thread, usage)	PR
MCP server	Model Context Protocol server for AI tools	registry
Telegram bot	Get the key, manage subscription, support	@ViBomemorybot

All projects share one license, one key, one product.

Dust is no longer dust. Dust is memory that was waiting for its hour.

Related projects

CloudArc — pack a 10 GiB folder into one .vibo archive at a flat ~26 MiB peak RSS, and read its index over HTTP without fetching the payload (Apache-2.0). Docs site
memory-shield — poisoning defense for agent memory (MIT).
ViBo-memory — persistent agent memory with L1/L2/L3 encryption.

vibo-memory

Documentation