Teams & Shared Memory.
One encrypted vault, many minds. Humans and their agents write into the same memory — institutional knowledge that actually compounds.
One Vault, Many Minds
A team vault is a shared encrypted memory that every member and every connected agent can read and write. A preference set by one agent becomes context for another. A decision recorded by a teammate becomes a constraint for the next session. Instead of each human hoarding their own notes in a private workspace, the team builds a single living codex — one that improves every week because every meeting, every review, every code change adds to it. Your agents don't start every conversation blank; they start with everything the team has already learned.
Shared memory for humans and agents.
Invite teammates. Register agents. Every write carries its author — institutional knowledge that compounds across every session.
One vault, many minds
Humans and agents write into the same encrypted codex. Knowledge compounds across every session.
Every memory attributed
Click any fact to see the agent, session, and human reviewer behind it. Provenance is always visible.
Role-scoped access
Viewer, Writer, Admin — per human and per agent. Revoke in one click; keys stop working immediately.
Invite humans, register agents
Invite teammates by email with role-scoped links (viewer, writer, admin) that expire. Register team agents — Claude Code for the backend engineer, Cursor for the frontend lead, a research agent that runs weekly sweeps — with per-agent API keys scoped to the vault. Every write is tagged with who did it (human or agent) and why, so the provenance is clear. Revoke an agent or remove a member at any time; their key stops working immediately and no further reads or writes reach the vault.
Agent Attribution Every Memory Can Explain
Every memory carries its origin: the agent that wrote it, the session it came from, the message that prompted it, and the human reviewer (if any) who confirmed it. Click any fact and see the chain. Was this decided by claude-code in a pair-programming session last Tuesday? Or did a teammate write it manually? Was it superseded by a later correction? The codex remembers not just the conclusions but the path that led there — essential for audit, debugging, and trust.
Role-Scoped Access
Vaults support three roles out of the box: Viewer (read-only search and recall), Writer (read + create memories + notes + media), Admin (manage members, invites, agents, and vault settings). Role changes take effect instantly. Agents can be scoped more tightly than humans — a visitor agent might get read-only access to a specific folder, while a dedicated research agent gets full write access to its own memory type but read-only to others. Privacy is the default; every access path is explicit.
Institutional Knowledge That Compounds
The longer your team uses ExoVault, the more valuable it gets. Onboarding a new engineer means giving them a connected agent that already knows your architecture, your conventions, your past decisions. A scheduled pipeline can run a weekly digest of what every agent learned. A new agent joins the vault and inherits everything — no knowledge transfer meetings, no stale wikis, no 'oh I forgot to tell you about that'. One vault, one memory, every mind aligned.
A codex worth keeping.
Free to start. Encrypted always. Connect your first agent in under a minute.