About wisegrid-mcp
wisegrid-mcp lets AI agents read and write WiseGrid sheets and run reports from a conversation. It supports paginated reads, bulk operations, and version-checked updates; an API key is required to access real data.
Key features
- 01List sheets and inspect columns and rows
- 02Page through sheets and records
- 03Insert records in bulk
- 04Update one or many records
- 05Delete records and their subtrees
- 06List and run reports
Use cases
- 01Look up sheet data in a conversation
- 02Add or edit records in bulk
- 03Apply version-checked row updates
- 04Run reports and retrieve their results
MCP support
The MCP server provides compatible clients with tools to read and write WiseGrid sheets, rows, columns, and reports. An API key is required for real data, and its scopes determine accessible content.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
uvx wisegrid-mcpWISEGRID_API_KEYSecretWiseGrid API key (create one under Settings)WISEGRID_BASE_URLOverride the API base URL for self-hosted instancesLocal servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.
Tools
12Saved tool list · updated 2w ago
This tool takes no parameters.
Details
From the project READMEWiseGrid MCP
Read and write WiseGrid sheets from Claude, Cursor, or any MCP client. List sheets, page through rows, add and update records, and run reports — from inside a conversation.
Unlike the other servers in this family, these tools operate on your real data, so an API key is required.
Install
{
"mcpServers": {
"wisegrid": {
"command": "uvx",
"args": ["wisegrid-mcp"],
"env": {
"WISEGRID_API_KEY": "wg_live_xxx"
}
}
}
}
Or pip install wisegrid-mcp.
| Variable | Required | Purpose |
|---|---|---|
WISEGRID_API_KEY | yes | Create one in WiseGrid under Settings |
WISEGRID_BASE_URL | no | Override for self-hosted instances |
The key's scopes decide what these tools can reach. Grant read-only scopes if you only want the agent to look.
Tools
| Tool | What it does |
|---|---|
whoami | The key's identity, scopes and rate limit — call this first when debugging |
list_sheets | Sheets the key can see, cursor-paginated |
get_sheet | Sheet + columns + first page of rows, in one call |
list_columns | Column IDs, which cell dictionaries are keyed by |
list_rows | Rows, cursor-paginated; each carries the version you need to update it |
add_rows | Bulk insert, partial success, max 2000 |
update_row | Update one row's changed cells, version-checked |
update_rows | Bulk update, version-checked per row, max 2000 |
delete_row | Delete a row and its subtree — requires confirm=True |
list_reports / get_report / run_report | Reports and their rows |
Things that will bite you otherwise
Cells are keyed by column ID, not column name. {"101": "Acme Corp"}, not
{"Client": "Acme Corp"}. Call get_sheet or list_columns first.
Updates are version-checked. Every row carries a version; you pass the one
you read, and if the row changed in the meantime you get version_conflict
instead of a silent overwrite. That is a normal outcome to handle — re-read the
row, decide whether your change still applies, then retry. Blindly bumping the
version to force the write is how concurrent edits disappear.
update_row merges, update_rows overwrites. The single-row tool takes
only the cells you changed. The bulk tool replaces the row's cell set, so send
the full set you want the row to end up with.
delete_row takes the children too. It removes the row and its entire
subtree, it is not undoable through the API, and it requires confirm=True
for exactly that reason. Read the rows first.
Writes are idempotent. Each carries an Idempotency-Key, so a retry after a
timeout cannot double-apply. Rate limits are honoured via Retry-After.
Development
git clone https://github.com/RyanKramer/wisegrid-mcp
cd wisegrid-mcp
pip install -e .
WISEGRID_API_KEY=wg_live_xxx wisegrid-mcp
About
Built by WiseGrid. Full REST API reference at wisegrid.co/docs/api.
MIT licensed.
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.