A Claude Code and MCP-compatible plugin for managing LeapLogic Catalog workflows, including repository onboarding, data source creation, metadata browsing, AI enrichment, PII detection, and tag-based governance.
- Browse Catalog — List repositories, data sources, and entities; explore entity details (columns, types, tags, indexes)
- Create Repository — Guided workflow: type selection → field discovery → verify → create
- Create Data Source — Guided workflow: repo selection → fields → verify → create → entities
- Verify Connection — Test connectivity for repos or data sources before creating them
- Enrich Entities — Submit an AI job to generate descriptions and detect PII for selected tables (or all tables in a data source)
- Manage Tags — Add tags to entities (tables/views) or attributes (columns) to organize and classify data assets
### Currently Supported Repository Types
The plugin currently supports two repository categories:
- **Cloud** — Snowflake
- **EDW** — Teradata- LeapLogic MCP server running with Keycloak OAuth configured
- A Dev or Full seat on a LeapLogic plan
The MCP server implements the MCP OAuth specification. When you connect:
- Your MCP client connects to the server
- Server returns
401 Unauthorized(no valid token yet) - Client discovers OAuth metadata at
/.well-known/oauth-authorization-server - Client opens the Keycloak login page in your browser automatically
- You log in with your LeapLogic credentials
- Client receives an access token and sends it with all subsequent requests
- Server auto-generates IDWSSOTOKEN internally — no manual token management required
The recommended way to set up the LeapLogic MCP server in Claude Code is by installing the plugin:
claude plugin install leaplogic-catalog@claude-communityManual setup:
claude mcp add leaplogic-metadata --transport http https://test.leaplogic.io/mcpAdd to your claude_desktop_config.json:
{
"mcpServers": {
"leaplogic-metadata": {
"url": "https://test.leaplogic.io/mcp"
}
}
}The server will trigger the Keycloak login flow automatically.
Install the plugin by typing the following command in Cursor's agent chat:
/add-plugin leaplogic-catalog
- Use the shortcut
Ctrl+Shift+Pto search forMCP: Add Server. - Select
HTTP. - Paste the server URL
https://test.leaplogic.io/mcpin the search bar. Then hitEnter. - When prompted for a server ID, enter
leaplogic-metadata. - Select whether to add globally or for the current workspace only.
Any editor that supports Streamable HTTP can connect:
{
"mcpServers": {
"leaplogic-metadata": {
"url": "https://test.leaplogic.io/mcp"
}
}
}| Setting | Required | Description |
|---|---|---|
mcp_server_url |
Yes | MCP server endpoint (e.g., https://test.leaplogic.io/mcp) |
That's it — no tokens to manage. Authentication is handled automatically via OAuth.
agent-toolkit/
├── .claude-plugin/
│ └── plugin.json # Plugin manifest
├── .cursor-plugin/
│ └── plugin.json # Cursor plugin manifest
├── .mcp.json # MCP server configuration (HTTP type)
├── server.json # Optional MCP server metadata using the official server schema
├── settings.json # Optional project metadata for internal/default agent selection
├── LICENSE # Apache-2.0
├── README.md # This file
├── assets/
│ └── leaplogic-icon.svg # Plugin icon
├── agents/
│ └── metadata-catalog.md # Specialist agent definition
└── skills/
├── shared/
│ └── SKILL.md # Shared input binding contract
├── browse-catalog/
├── create-repository/
├── create-datasource/
├── enrich-entities/
├── manage-tags/
└── verify-connection/
| Tool | Description |
|---|---|
get_repository_types |
List supported repository types by category |
get_source_fields |
Discover required fields for a repo/DS type |
get_repositories |
List all registered repositories |
get_datasources |
List data sources in a repository |
verify_repository |
Test repository connectivity |
create_repository |
Register a new repository |
verify_datasource |
Test data source connectivity |
create_datasource |
Add a data source to a repository |
get_datasource_entities |
List entities (tables/views) in a data source |
get_sync_status |
Check import/sync progress of data sources |
get_entity_details |
Get column-level details, tags, and indexes for an entity |
add_tags |
Add tags to an entity (table/view) or attribute (column) |
enrich_entities |
Submit an AI job that generates descriptions and detects PII for selected entities |
- Name entities semantically — Use meaningful names for repositories, data sources, and schemas (e.g.,
sales-warehousenotrepo_01) - Tag everything — Tags improve discoverability; apply domain, sensitivity, and ownership tags early
- Enrich early — Run
enrich_entitiesright after data source creation to populate descriptions and PII labels while context is fresh
- Be specific about scope: "Show me all PII columns in the customer_orders table"
- Use catalog hierarchy: "List data sources in the production-warehouse repository"
- Chain operations: "Create a Snowflake repository, add a data source, then enrich all discovered tables"
- Always verify before creating — The plugin enforces verify → create ordering. Connection issues are cheaper to fix before resource creation.
- Browse first — Explore the catalog before making changes to understand existing structure.
- Bulk enrich — Use "enrich all entities" for newly imported data sources rather than enriching one at a time.
| Problem | Solution |
|---|---|
| Login page doesn't appear | Ensure Keycloak is reachable at configured host:port |
| "Connection refused" | Verify MCP server host/port and network access |
| Tools not loading | Confirm the MCP server is reachable and exposing the metadata catalog tools |
| Token expired mid-session | MCP client will auto-refresh; if not, reconnect |
| Skills not appearing | Run /reload-plugins in Claude Code or restart the client |