KiviDB from AI assistants
KiviDB Cloud ships an MCP server, so Claude, ChatGPT, Gemini and any other assistant that speaks the Model Context Protocol can run your databases, query…
KiviDB Cloud ships an MCP server, so Claude, ChatGPT, Gemini and any other assistant that speaks the Model Context Protocol can run your databases, query them, and pass context to each other through one.
It is published on npm as @kividbio/mcp-server. There is nothing to
install: the clients start it on demand.
The API key is the limit
Everything the server does, it does as an API key. The key belongs to one organization and carries a role, so what an assistant can see and what it may change are decided by KiviDB, not by the assistant and not by this package.
Mint one in the console under API keys. Give it the smallest role that does the job:
| Role | Can |
|---|---|
member | Create, change and delete databases; read and write data |
admin | The same, plus mint and revoke API keys |
owner | The same, plus billing and membership |
When KiviDB refuses something, the assistant is told why in KiviDB's own words — "free tier instances are one per person", not "request failed" — so it can pass that on instead of retrying.
Revoking the key in the console stops every assistant holding it, immediately.
What it can do
Databases. List them, read their detail, create one (including a free-tier
database that costs nothing), resize the disk, scale replicas, upgrade
Essentials to Pro, and delete. A database can also be created in one of your
organization's own connected cloud accounts (list_cloud_accounts); connecting
an account stays in the console, with the owner.
Creating or upgrading a paid database costs money, and the tools say so, so an assistant asks before spending. Deleting requires the database's exact current name, which the server checks against KiviDB before it deletes anything — an assistant that has confused two ids is stopped rather than believed.
Shared context. remember writes a note; recall reads it back. Notes are
stored in your own database, so any assistant pointed at the same database and
namespace sees them. That is what lets one assistant pick up where another left
off, and it needs no agreement between them beyond a name.
Vectors. vector_search finds the nearest neighbours of an element in a
vector set. KiviDB does not build embeddings: the vectors have to be in the
database already, added with VADD.
Backups. Take a disk snapshot, and list the ones you have.
Anything else. run_command runs a single KiviDB command — SET, GET,
FT.SEARCH, VADD, VSIM, and the rest. Commands that manage the server
itself are refused by KiviDB, not by this package.
Sharing context between assistants
Point two assistants at the same database and the same namespace, and they read each other's notes:
In ChatGPT: Remember that we moved the staging allowlist into Terraform.
In Claude, later: What do we know about the allowlist?
[mul8o6rx] 2026-09-28 by chatgpt (staging, deploy)The staging allowlist reverts on every deploy until the Terraform apply runs.
Notes carry an author and optional tags, and recall filters on both, so you
can ask what one assistant concluded without the others' notes crowding it out.
Namespaces keep projects apart. The default is shared; pass a different
namespace and the notes are invisible to anyone using another one.
Nothing is remembered unless an assistant is asked to remember it. These are notes in your database, not a transcript.
What it costs
The server itself costs nothing. What it does can: creating a paid database, resizing one, or upgrading to Pro all bill the organization that owns the key. A free-tier database costs nothing and is a good place to start.
Next
- Connect your assistant — Claude, ChatGPT, Gemini, Cursor and VS Code, and how to run it as a shared HTTPS service.