The Muvia public API lets an integration read its own company's Muvia data: the company, its projects, the datasets of each project and the rows they hold, and the nodes that collect the data. Version 1 is read-only.
Every request carries an API key. A key belongs to one company and only ever sees that company's data.
Building with an AI agent? Start with the Agent quickstart: the whole path, REST and MCP, with complete requests and responses.
The base URL is https://api.muviabi.com/api/public, the server shown at the top
of the API Reference.
Every v1 path starts with /v1, so the company's projects are at
https://api.muviabi.com/api/public/v1/projects.
A company administrator issues keys in Muvia, under
Settings → Company → API keys. A key has a name, a lifetime in days (90
by default, at most 3650) and the scopes the integration
needs — for example settings_read to list projects and datasets_read to
read datasets.
The key, mvk_<key_id>_<secret>, is shown once, when it is issued. Store
it in a secret manager. A lost key cannot be recovered: revoke it and issue a
new one.
export MUVIA_API_KEY="mvk_<key_id>_<secret>"
curl -H "X-Api-Key: $MUVIA_API_KEY" https://api.muviabi.com/api/public/v1/whoami
GET /v1/whoami answers with the key (its scopes and expiry), the company, the
request limits, whether the MCP server is available and
where the documentation is. Any valid key may call it, and it does not count
toward the daily quota. Then list the projects:
curl -H "X-Api-Key: $MUVIA_API_KEY" https://api.muviabi.com/api/public/v1/projects
The answer is one page of the company's projects:
{
"items": [
{
"id": "66f1a2b3c4d5e6f708192a3b",
"name": "Line 3 monitoring",
"description": "Sensors on the third production line",
"is_default": true,
"created_at": "2026-05-12T08:30:00Z",
"updated_at": "2026-09-20T14:05:12Z"
}
],
"total": 1,
"page": 1,
"size": 50,
"pages": 1
}
From a project, read its datasets, then a dataset's rows:
curl -H "X-Api-Key: $MUVIA_API_KEY" \
https://api.muviabi.com/api/public/v1/projects/<project_id>/datasets
curl -H "X-Api-Key: $MUVIA_API_KEY" \
"https://api.muviabi.com/api/public/v1/projects/<project_id>/datasets/<dataset_id>/rows?size=100"
The API Reference has the full schema of every resource.
Every error answers with the same envelope:
{
"error": {
"code": "unauthorized",
"message": "Invalid or missing API key",
"status": 401,
"hint": "Send a valid key in the X-Api-Key header; check it has not expired or been revoked in Settings > API keys, and that it belongs to this environment."
}
}
Branch on code, never on message. The Errors guide lists
every code.
Every REST operation, with its effect, whether it spends the company's daily
quota, and the scope it needs. The OpenAPI document states the same on each
operation as x-muvia-effect and x-muvia-quota.
| Method | Path | Effect | Counts toward the daily quota | Scope |
|---|---|---|---|---|
| GET | /v1/whoami | read | no | none (any valid key) |
| GET | /v1/company | read | yes | settings_read |
| GET | /v1/projects | read | yes | settings_read |
| GET | /v1/projects/{project_id} | read | yes | settings_read |
| GET | /v1/projects/{project_id}/datasets | read | yes | datasets_read |
| GET | /v1/projects/{project_id}/datasets/{dataset_id} | read | yes | datasets_read |
| GET | /v1/projects/{project_id}/datasets/{dataset_id}/rows | read | yes | datasets_read |
| GET | /v1/projects/{project_id}/nodes | read | yes | nodes_read |
| GET | /v1/projects/{project_id}/nodes/{node_id} | read | yes | nodes_read |
Version 1 of the REST API is read-only. The MCP server
is where an agent can also write: each MCP tool carries an annotation that says
whether it reads (readOnlyHint) or writes.
The datasets of a project are the project's own datasets plus the physical datasets of the project's database.