A Claude Code plugin that helps dbt users quickly onboard to Recce - the open-source data validation and diff tool for dbt.
- Guided Setup -
/recce-setupwalks you through environment configuration - PR Analysis -
/recce-pranalyzes data impact of pull requests - Data Checks -
/recce-checkruns validation checks between environments - CI/CD Setup -
/recce-cigenerates GitHub Actions workflows for Recce Cloud
Step 1: Add the Recce marketplace to Claude Code
In Claude Code, run:
/plugin marketplace add DataRecce/recce-claude-plugin
Step 2: Install the plugin
/plugin install recce-quickstart@recce-claude-plugin
Or use the interactive installer:
/plugin
Then navigate to Discover tab, find recce-quickstart, and press Enter to install.
Step 1: Clone the repository
git clone https://github.com/DataRecce/recce-claude-plugin.git
cd recce-claude-pluginStep 2: Add as local marketplace
In Claude Code, run:
/plugin marketplace add /path/to/recce-claude-plugin
Step 3: Install the plugin
/plugin install recce-quickstart@recce-claude-plugin
You can install the plugin at different scopes:
| Scope | Command | Description |
|---|---|---|
| User (default) | /plugin install ... |
Available across all your projects |
| Project | /plugin install ... --scope project |
Shared with team via .claude/settings.json |
| Local | /plugin install ... --scope local |
Only for current repository, not shared |
After installation, verify the plugin is working:
/plugin
Navigate to the Installed tab to see recce-quickstart.
- Navigate to your dbt project directory
- Run
/recce-setupto configure your environment - Use
/recce-pror/recce-checkto analyze data changes
| Command | Description |
|---|---|
/recce-setup |
Guided environment setup (installs dependencies, generates artifacts, starts MCP server) |
/recce-pr [url] |
Analyze PR data changes (auto-detects PR from current branch) |
/recce-check [type] [selector] |
Run data validation checks (row-count, schema, profile, query-diff) |
/recce-ci |
Set up Recce Cloud CI/CD for GitHub Actions (PR review + main branch workflows) |
- Python 3.8+
- dbt (any adapter: duckdb, postgres, bigquery, snowflake, etc.)
- Git
The plugin will guide you to install these if missing:
pip install recce- Recce CLIpip install 'recce[mcp]'- Recce MCP Server (for AI-powered analysis)
This plugin:
- Detects when you're in a dbt project (via
dbt_project.yml) - Guides you to set up base and current dbt artifacts
- Starts a Recce MCP server for AI-powered analysis
- Provides Claude with tools to analyze data changes
When the Recce MCP server is running, Claude has access to these tools:
| Tool | Description |
|---|---|
lineage_diff |
Compare data lineage between environments |
schema_diff |
Detect schema changes (columns, types) |
row_count_diff |
Compare row counts between environments |
profile_diff |
Statistical profiling comparison |
query_diff |
Run custom SQL queries for comparison |
Disable the plugin:
/plugin disable recce-quickstart@recce-claude-plugin
Re-enable the plugin:
/plugin enable recce-quickstart@recce-claude-plugin
Uninstall the plugin:
/plugin uninstall recce-quickstart@recce-claude-plugin
Update marketplace:
/plugin marketplace update recce-claude-plugin
Want to automate data validation in CI/CD? Recce Cloud offers:
- Automatic PR analysis
- Data quality gates
- Team collaboration
- Historical tracking
- Verify Claude Code version is 1.0.33 or higher:
claude --version - Check plugin is installed:
/plugin→ Installed tab - Check for errors:
/plugin→ Errors tab
- Ensure you're in a dbt project directory (has
dbt_project.yml) - Verify Recce is installed:
pip install 'recce[mcp]' - Check if port 8081 is available (or set
RECCE_MCP_PORT=8085) - Run the setup command:
/recce-setup
- Ensure plugin is enabled:
/plugin→ Installed tab → check status - Restart Claude Code to reload plugins
MIT