Monte Carlo's unified agent toolkit plugin for the Cursor editor. Delivers data observability skills and enforcement hooks as named features within a single plugin.
| Feature | Description |
|---|---|
| MC Prevent | Detect and prevent breaking schema changes using Monte Carlo lineage and monitoring data. |
Requires Python 3.10+.
bash <(curl -fsSL https://raw.githubusercontent.com/monte-carlo-data/mc-agent-toolkit/main/plugins/cursor/scripts/install.sh)-
Clone the repository:
git clone https://github.com/monte-carlo-data/mc-agent-toolkit.git cd mc-agent-toolkit -
Run the install script:
bash plugins/cursor/scripts/install.sh
This copies the plugin (with symlinks resolved) to
~/.cursor/plugins/local/mc-agent-toolkit. -
Restart Cursor or run Developer: Reload Window from the Command Palette (
Cmd+Shift+P). -
The Monte Carlo MCP server will prompt for OAuth authentication on first use.
- Hook denials may not be enforced. Cursor's
beforeReadFile(and potentially otherbefore*hooks) may fail to block the operation even when the hook exits with a deny response. This means Prevent hooks can detect and warn about breaking changes but cannot guarantee the edit is stopped. See Cursor forum discussion for details.
The toolkit plugin wraps shared skills and hook logic, with each feature namespaced independently:
- Skills — symlinked from
skills/at the repo root (shared across all editors) - Shared hook logic — copied from
plugins/shared/prevent/lib/into each plugin'shooks/prevent/lib/(platform-agnostic business logic) - Adapter hooks — Cursor-specific JSON parsing and output formatting under
hooks/prevent/ - MCP config — Monte Carlo MCP server connection
See the plugins README for the overall plugin architecture and editor support comparison.