What MemVault is

MemVault is a universal, vendor-neutral floating memory layer for your AI agents. While SkillVault gives your agents a home for what they know how to do (skills), MemVault gives them a home for what they remember: your project decisions, brand rules, past bug investigations, and client constraints.

MemVault does not belong to OpenAI, Anthropic, Google, or any single tool vendor. It is an independent, user-owned memory vault reachable from everywhere you interact with AI: in your terminal via Claude Code, in your editor via Cursor or Windsurf, in your browser on ChatGPT.com or Claude.ai, or on your phone talking to Kimi, Gemini, or Claude mobile.

One persistent memory graph. Every platform, every device, every account.

The problem we are solving: The Memory Walled Garden

Agent memory has been trapped inside closed silos. OpenAI has memory inside ChatGPT. Claude Code has memory inside local project files. Cursor has its own workspace notes. Every vendor builds their own island.

The real problem is the total lack of memory across the tools you actually use:

Walled Gardens: Context created inside ChatGPT on the web stays trapped inside ChatGPT. It cannot inform Claude Code in your terminal, and Cursor has no idea it exists.
Cross-Device Amnesia: If you brainstorm a business model, pricing cap, or technical decision with Claude or Kimi on your phone during your commute, that context is dead the moment you open your laptop. You have to explain it all over again.
Tool Disconnect: When you switch from Cursor to Windsurf, or from Claude Code to DeepSeek Harness, your history resets to zero. Your agent has amnesia simply because you switched windows.
The Copy-Paste Tax: Because AI tools do not share state, you become an exhausted bridge. You spend your day re-typing rules, guidelines, and past fixes into different chat windows.
Vendor Lock-in: Storing your context inside a closed system locks you to that subscription. When a better model comes out tomorrow, switching means abandoning months of project context.

MemVault turns memory into an open utility. It is not tied to one model. It is a floating memory layer that travels with you across accounts, platforms, and devices.

How cross-platform memory works: One brain, every surface

Here is what daily work looks like with MemVault:

Morning on Mobile (Kimi / Claude iOS / ChatGPT): On your commute, you decide on a pricing rule and outreach tone on your phone:

> "We have a hard 10% maximum discount cap on annual plans, and we never use corporate hype in customer communications."

Because your mobile assistant is connected to your MemVault URL, that decision is captured into your universal vault in real time.

At Your Desk (Cursor / Claude Code / DeepSeek Harness): You sit down at your workstation, launch Cursor or Claude Code, and prompt: "Draft the onboarding sequence and checkout handler." Your agent performs a pre-flight recall against MemVault. It retrieves your 10% discount cap and tone rules set on your phone an hour earlier, and implements them immediately without you ever typing the requirement twice.
In the Terminal (Git Milestones): Later, you commit code in your terminal. The skillvault git hook automatically captures your commit summary, diff, and author into MemVault as a durable milestone.
In the Web Browser (ChatGPT.com / Claude.ai): Days later, a teammate asks ChatGPT about your pricing rules. ChatGPT queries your vault URL and recites the exact 10% discount cap and the milestone where it was confirmed.

The Four Delivery Surfaces

MemVault delivers your memory across four interoperable surfaces:

1. Authenticated Vault URLs (/v/<user>/t/<token>): For web chats and mobile AI assistants (ChatGPT.com, Kimi, Claude.ai, mobile apps) that can fetch URLs. Returns clean Markdown formatted for LLM context windows.
2. OpenAPI and REST Connectors: Connect directly to platforms like Perplexity Pro via the OpenAPI specification (/api/openapi.json), or custom agent harnesses via Bearer auth.
3. Model Context Protocol (MCP): Connects natively into Cursor, Windsurf, Claude Desktop, and ChatGPT Custom Connectors as first-class tools (memvault_recall, memvault_ingest).
4. Local CLI Tool (skillvault mem): Fast terminal commands for Claude Code, DeepSeek Harness, Aider, OpenCode, Codex, and local automation scripts.

Intelligent Cognitive Architecture: Beyond Dumb Text Dumps

Memory needs precision. Dumping thousands of words of conversational transcripts into every prompt wastes tokens, increases cost, and distracts the model.

MemVault works like a cognitive coprocessor:

1. Vector Semantic and Graph Indexing

MemVault indexes memories using dense semantic embeddings paired with associative graphs. When your agent asks about your discount policy, it finds your rule even if you phrased the question differently.

2. Ebbinghaus Retention Dynamics

Human memory prioritizes recent, reinforced thoughts over old noise. MemVault applies an Ebbinghaus retention curve: temporary debugging notes fade naturally over time unless reinforced through repeated recall.

3. Pinned Golden Rules (Strength 1.0)

Critical brand standards, security rules, and user preferences can be pinned. Pinned memories never decay. Their strength remains locked at 1.0 forever, making sure core invariants are never forgotten.

4. Peripheral Attention Engineering (PAE)

MemVault organizes recalled context into disciplined, token-budgeted slots:

[ASKER CONTEXT]: Pinned user preferences and immutable project guardrails.
[SITUATIONAL CONTEXT]: Active branch context, recent milestones, and repository trajectory.
[REFERENCE KNOWLEDGE]: Semantically relevant past solutions and bug root causes.

PAE ensures memory never consumes more than a calibrated fraction of your context window, leaving maximum room for your actual work.

Team & Organization Governance

Memory fragmentation gets worse on teams. When one person solves an obscure bug, establishes a client rule, or configures a deployment pipeline, that knowledge stays trapped on their laptop.

MemVault bridges the entire team with a dual-layered design:

Personal Vault Layer: Individual team members keep private notes, scratchpads, and personal preferences that never leak into the shared company space.
Shared Team Vault Layer: Team repositories automatically pool project milestones, architectural decisions, and bug fixes into a centralized organization memory graph.
📌 Pinned Org Standards: Leads pin golden rules (such as "All public endpoints require authentication before touching the database" or "Outbound client proposals must cap discounts at 10%"). These standards stay locked at strength 1.0 and inject into the prompt context of every team member and agent.
Contributor Attribution: Every team memory records who created it, so your team always knows where a decision came from.
Security Controls: Repository blacklists, project-level exclusions, and custom retention windows keep your data compliant with internal policies.

Data Ownership & Vendor Independence

When your agent memory lives inside a closed ecosystem, you are locked in. If a provider changes their terms, limits your rate, or falls behind a new model, you cannot take your context with you.

With MemVault:

You Own Your Context: Your memory lives in your encrypted vault, not inside a closed model subscription.
Zero Model Training: We never use your memories, commit diffs, or project notes to train public or proprietary models.
Switch Models Freely: Move between Claude, GPT, DeepSeek, or local open-weight models without losing a single decision, preference, or milestone.

Pricing

MemVault is built into all SkillVault subscription tiers:

Free: 50 persistent memories, 3 skills, 100 MB storage. Great for trying universal memory on personal projects.
Pro: €8.95/month. 5,000 persistent memories, 50 skills, 2 GB storage, automatic git commit capture. Built for everyday multi-agent workflows.
Power: €24.95/month. 50,000 persistent memories, unlimited skills, 15 GB storage, neural reranking, and associative graph traversal.
Team & Enterprise: €14.95/seat/month. 100,000 memories per organization, shared team vaults, Pinned Org Standards, role-based access, and compliance controls.

How to start

Get started in under two minutes:

Create your account at /auth/signup and copy your vault address (https://app.skillvault.dev/v/<username>).
For coding agents (Claude Code, Cursor, Windsurf), tell your agent: Set up my skillvault https://app.skillvault.dev/v/<username>
For web chats and mobile apps (ChatGPT, Claude.ai), paste your private vault URL into your conversation.

Get in touch

Questions about MemVault, enterprise setups, or custom team integrations? Email us anytime at support@skillvault.dev.