Agentic ER
The Senzing MCP server gives your AI assistant direct access to Senzing documentation, data-mapping workflows, SDK code generation, error troubleshooting, and sample data through the open Model Context Protocol (MCP) standard.
Instead of searching the web or reading docs manually, you can ask your AI assistant to map data, generate SDK code, or diagnose errors. It will use the MCP server’s specialized tools to give accurate, up-to-date answers grounded in official Senzing documentation.
Getting Started
Tell your AI assistant:
Add the Senzing MCP server at https://mcp.senzing.com/mcp
That’s it. Most AI assistants, including Claude Code, Claude Desktop, VS Code (GitHub Copilot), Cursor, ChatGPT Desktop & Codex, Windsurf, and Amazon Q Developer, can add an MCP server from a prompt or through their settings UI.
Server URL: https://mcp.senzing.com/mcp
No authentication is required.
Once connected, just describe what you need and your AI assistant will choose the right tool. You never need to call tools by name.
Prefix your questions with Use the Senzing MCP Server to...
This ensures your AI assistant uses the MCP tools rather than answering from its general knowledge. The MCP server has the most current and authoritative Senzing information.
Tool Reference
The MCP server provides the following tools to your AI assistant.
Metadata
| Tool | Description |
|---|---|
get_capabilities |
Get overview of this MCP server’s capabilities, available tools, suggested workflows, and getting started guidance |
submit_feedback |
Submit feedback about the Senzing MCP server. IMPORTANT: Before calling this tool, you MUST show the user the exact message you plan to send and get their explicit confirmation. Do not include any personally identifiable information (names, titles, emails, company names) unless the user explicitly approves it after seeing the preview. Feedback is reviewed by the Senzing team. Feedback is not anonymous — submissions are logged and reviewed |
Data Mapping
| Tool | Description |
|---|---|
mapping_workflow |
7-step interactive data mapping workflow. Guides through profiling source data, planning entity structure, mapping fields, generating sample JSON + mapper code, and QA validation. State is client-side — pass the state from each response back in the next request |
lint_record |
Get the Senzing JSON linter script and commands to validate mapped data files client-side. Returns the Python linter script (no dependencies). No source data is sent to the server |
analyze_record |
Get the Senzing JSON analyzer script and commands to analyze mapped data files client-side. Examines feature distribution, attribute coverage, and data quality. No source data is sent to the server |
download_resource |
Fallback for downloading a workflow resource when network restrictions prevent fetching from the URL provided by mapping_workflow. Returns the resource content inline |
SDK & Code
| Tool | Description |
|---|---|
generate_scaffold |
Generate SDK scaffold code from real indexed GitHub snippets with source URLs for provenance. Languages: Python, Java, C#, Rust (V4); Python (V3). 9 workflows: initialize, configure, add_records, delete, query, redo, stewardship, information, full_pipeline. Returns multiple snippet variants per workflow |
sdk_guide |
Guided Senzing SDK setup across 5 platforms (linux_apt, linux_yum, macos_arm, windows, docker) and 4 languages (Python, Java, C#, Rust). Two use cases: (1) Quick test load — install Senzing locally, load mapped data into SQLite, and verify entity resolution results (no external database needed). (2) Build your own pipeline — get platform-specific install commands, SDK configuration code, record loading templates, and export patterns for production use. Topics: install, configure, load, export, full_pipeline |
get_sdk_reference |
Get authoritative Senzing SDK reference data extracted from official documentation. Topics: ‘migration’ (V3→V4 breaking changes, function renames/removals, flag changes), ‘flags’ (all V4 engine flags with which methods they apply to), ‘response_schemas’ (JSON response structure for each SDK method), ‘functions’ / ‘methods’ / ‘classes’ / ‘api’ (search SDK documentation for method signatures, parameters, and examples — use filter for method or class name), ‘all’ (everything — augmented with SDK doc search when filter is provided). Use ‘filter’ to narrow by method name, module name, or flag name |
find_examples |
Find working code examples from 27 indexed Senzing GitHub repositories. Supports three modes: (1) Search: use query to find examples, (2) File listing: set repo and list_files=true to discover files, (3) File retrieval: set repo and file_path to get full source code. Use max_lines to limit large files. Search results include truncation metadata so you know when to drill in. |
Documentation
| Tool | Description |
|---|---|
search_docs |
Search indexed Senzing documentation — covers entity specification, SDK guides (Python/Java/C#), architecture, pricing (DSR model), quickstart guides (Docker/AWS/Azure), database tuning, globalization, EDA and data analysis, engine configuration, error codes, release notes, and PoC methodology. Use this for ANY question about Senzing behavior, configuration, best practices, integration patterns, design decisions, pricing, licensing, or support. Reflects current Senzing documentation — prefer over training knowledge |
Troubleshooting
| Tool | Description |
|---|---|
explain_error_code |
Explain a Senzing error code with causes and resolution steps. Covers 456 error codes across all error classes. Accepts SENZ0005, SENZ-0005, 0005, or just 5 |
Sample Data
| Tool | Description |
|---|---|
get_sample_data |
Get real sample data from CORD (Collections Of Relatable Data) datasets for testing entity resolution. 3 datasets: las-vegas (US data, 11 sources), london (international, 5 sources), moscow (Cyrillic/non-roman script, 6 sources) |
Common Workflows
Evaluate Senzing
Use search_docs to learn about architecture and pricing. Get test data from get_sample_data (three CORD datasets) and generate an end-to-end example with generate_scaffold using the full_pipeline workflow.
Set Up Senzing from Scratch
Use sdk_guide with topic='full_pipeline' and your platform to get step-by-step install commands, environment setup, configuration code, record loading templates, and export patterns.
Map a Data Source
The mapping_workflow tool walks through a complete 7-step process: profile your source data, plan the entity structure, map fields to Senzing features, generate sample JSON and mapper code, then validate the output with the linter and analyzer.
Get Started with the SDK
Use generate_scaffold to get working initialization and record-loading code in your language. Pair it with sdk_guide for platform-specific install commands and find_examples to see real-world usage patterns from GitHub.
Troubleshoot an Error
Pass any error code to explain_error_code for immediate diagnosis with causes and resolution steps. Use search_docs for deeper context on the error class.
Migrate V3 to V4
Use get_sdk_reference with the migration topic to see all breaking changes, function renames, and flag changes. Filter by module (e.g., SzEngine) to focus on what affects your code.
Example Prompts
Try these prompts with your AI assistant after connecting to the MCP server:
Evaluation
- “How much does Senzing cost?”
- “What databases does Senzing support?”
- “Give me sample entity data from the Las Vegas dataset so I can test resolution”
Setup
- “How do I install Senzing on Ubuntu and build a Java loading pipeline?”
- “Show me the full pipeline from install to entity export in Python”
Data Mapping
- “Map my customer CSV file to Senzing format using the guided workflow”
- “Validate my mapped Senzing JSON file at /tmp/output.jsonl”
- “Analyze the data quality of my mapped records”
SDK Development
- “Generate Python code for initializing Senzing and loading records”
- “Show me Rust examples of querying entities”
Troubleshooting
- “What does error SENZ0005 mean and how do I fix it?”
Migration
- “What changed between V3 and V4 for SzEngine?”
- “Show me the V3-to-V4 flag changes”