Snowfakery MCP Server
Provides tools for generating Salesforce test data and mappings for CumulusCI workflows.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Snowfakery MCP Server Generate a Snowfakery recipe for 50 Account records with random names and phone numbers."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Snowfakery MCP Server
Power up your AI workflows with Snowfakery data generation — Use Claude, ChatGPT, and other AI assistants to author, debug, and run data recipes through the Model Context Protocol.
MCP Registry
mcp-name: io.github.composable-delivery/snowfakery-mcp
Related MCP server: Salesforce MCP
What is this?
Snowfakery is a YAML-based tool for programmatically generating test data. This MCP server connects Snowfakery to AI assistants, letting you:
Draft recipes with AI assistance backed by real Snowfakery docs and examples
Validate recipes before running them with detailed error feedback
Execute recipes and iterate on results interactively
Debug issues with static analysis and recipe inspection
Generate Salesforce mappings for CumulusCI workflows
Perfect for teams that need realistic test data—from Salesforce admins to developers building data pipelines.
Quick Start
Install uv
We recommend using uv for installs and for running from source.
Install
uv(macOS/Linux):curl -LsSf https://astral.sh/uv/install.sh | shInstall
uv(Windows PowerShell):powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"
See the official uv install docs: https://docs.astral.sh/uv/getting-started/installation/
Claude Desktop (recommended)
For Claude Desktop, prefer using the .mcpb bundle from Releases:
Download the latest
.mcpbfrom https://github.com/composable-delivery/snowfakery-mcp/releasesAdd the bundle in Claude Desktop as an MCP server bundle
This bundle includes the pinned runtime metadata (uv.lock, manifest.json) and is the easiest way to get a reproducible setup.
Install & Run (CLI)
# Recommended: isolated install
uv tool install snowfakery-mcp
# Then run the server
snowfakery-mcpOr from source:
git clone https://github.com/composable-delivery/snowfakery-mcp.git
cd snowfakery-mcp
uv sync
uv run snowfakery-mcpConnect to Claude (Desktop)
Add to your Claude Desktop claude_desktop_config.json:
{
"mcpServers": {
"snowfakery-mcp": {
"command": "snowfakery-mcp"
}
}
}Then ask Claude:
"Show me an example Snowfakery recipe" or "Help me write a recipe to generate 100 Salesforce accounts"
Features
Resources — Access docs, examples, and schemas:
Snowfakery documentation and recipe examples
JSON schema for recipe validation
Run outputs and artifacts
Tools — Interact with recipes:
Validate & analyze recipes (catch errors early)
Run recipes and capture output
List & retrieve example recipes
Generate CumulusCI mapping files
Learn More
MCP_SERVER_SPEC.md — detailed design and tool catalog
Snowfakery docs — recipe language reference
Contributing — how to contribute
Community
We want this to be welcoming at any level. Questions, ideas, and contributions are always welcome!
Questions & ideas? Open a GitHub Discussion
Found a bug? Open an Issue with a minimal recipe
Want to contribute? See CONTRIBUTING.md
Security concern? See SECURITY.md
Development
# Install dev dependencies
uv sync --all-groups
# Run tests
uv run pytest
# Type check
uv run mypy snowfakery_mcp
# Lint & format
uv run ruff check snowfakery_mcp tests scripts evals
uv run ruff format snowfakery_mcp tests scripts evalsEvals (Agentic Testing)
This repo includes inspect-ai tasks for testing the MCP server with AI models:
# Install eval dependencies
uv sync --group evals
# Run evaluation
uv run inspect eval evals/inspect_tasks.py@snowfakery_mcp_agentic --model openai/gpt-4o-miniSee evals/ for more examples and troubleshooting.
Notes
The repo includes the upstream Snowfakery repo as a git submodule (
Snowfakery/) for developmentWhen running from source, use
uv run ...to ensure the pinned environmentPyPI installs use bundled docs/examples (no submodule required)
Releases
See GitHub Releases for sdist, wheel, and .mcpb bundles (recommended for Claude Desktop).
Available Tools
10 toolsanalyze_recipeARead-onlyIdempotent
Parse and analyze a Snowfakery recipe structure.
Returns structural information about the recipe including:
Tables and their fields
Declared plugins and options
Random reference usage
Recipe version
Use this before running to understand recipe structure.
Inputs: recipe_path, recipe_text.
generate_mappingA
Generate a CumulusCI mapping.yml file from a Snowfakery recipe.
Creates the mapping file needed to load Snowfakery-generated data into Salesforce using CumulusCI. Returns a preview and artifact URI.
Inputs: recipe_path, recipe_text, load_declarations_paths. Returns: object
get_exampleARead-onlyIdempotent
Fetch a Snowfakery example recipe by name.
Returns the full text of the specified example recipe. Use list_examples first to see available examples.
Inputs: name (required). Returns: name, path, content
get_schemaARead-onlyIdempotent
Return the Snowfakery recipe JSON schema.
Use this schema to understand the structure of valid recipes and for validation purposes.
Inputs: no named parameters. Returns: object
iterative_recipe_genA
Create a recipe iteratively with validation.
This tool uses the LLM (via sampling) to draft a Snowfakery recipe, validates it, and if it fails, asks the LLM to fix it. Returns the final valid recipe or the last attempt.
Inputs: goal (required), max_iterations. Returns: result
list_capabilitiesARead-onlyIdempotent
Return Snowfakery + server capability info.
Use this tool first to understand what the server can do, including:
Snowfakery version
Supported output formats
Server limits (timeout, max output, max reps)
Available resources
Inputs: no named parameters. Returns: limits, resources, snowfakery_version, supported_output_formats
list_examplesARead-onlyIdempotent
List available Snowfakery example recipe files.
Returns a list of example recipe filenames from the bundled examples. Use prefix to filter results (e.g., "salesforce" for Salesforce examples).
Inputs: prefix. Returns: examples
run_recipeA
Run a Snowfakery recipe and generate fake data.
Executes the recipe and returns generated output along with artifact URIs.
The complete output is always written to disk and available via the
returned resource URI regardless of capture_output - that setting
only controls how much of it also comes back inline in this response.
Inputs: reps, options, recipe_path, recipe_text, +7 more. Returns: object
search_docsARead-onlyIdempotent
Search Snowfakery documentation for a query string.
Returns matching lines from the markdown documentation. Useful for finding specific syntax, features, or examples.
Inputs: query (required), limit. Returns: hits, query, truncated
validate_recipeARead-onlyIdempotent
Validate a Snowfakery recipe without generating data.
Checks recipe syntax and structure. Returns validation errors if any. Use either recipe_path (file on disk) or recipe_text (inline YAML).
Inputs: options, recipe_path, recipe_text, strict_mode, +2 more. Returns: valid, errors
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 10 tools
Most tools have distinct purposes, but analyze_recipe and validate_recipe could cause some confusion—both inspect recipes but one focuses on structure, the other on syntax. Overall, the differences are clear with careful reading.
All tool names follow a consistent verb_noun snake_case pattern (e.g., get_example, list_examples, run_recipe), making them predictable and easy to understand.
10 tools is well-scoped for a recipe-based data generation server, covering analysis, validation, execution, documentation, and examples without being overwhelming.
The tool surface covers the main workflow (create, validate, run, analyze) and includes helpful extras like documentation search and examples. A minor gap is the lack of a tool to edit existing recipes, but the iterative generation tool mitigates this.
Maintenance
Related MCP Connectors
Enable secure connectivity between Sentry issues and debugging data, and LLM clients, using a Model Context Protocol (MCP) server.
A Model Context Protocol server for Wix AI tools
A comprehensive Model Context Protocol (MCP) server that enables AI assistants to interact with yo…
Plan Salesforce deploys, open pull requests and trigger pipelines from your AI client.
Related MCP Servers
- FlicenseNot gradedqualityCmaintenanceEnables safe execution of Python code with AI assistance and integration with Model Context Protocol tools.1-
- FlicenseNot gradedqualityDmaintenanceEnables AI agents to query, mutate, and analyze Salesforce CRM data natively through the Model Context Protocol, without requiring API glue code.6 npm3-
- AlicenseNot gradedqualityDmaintenanceEnables AI assistants to interact with Databricks workspaces, running SQL queries, managing jobs, and exploring schemas via the Model Context Protocol.1GPL 3.0
- AlicenseNot gradedqualityDmaintenanceExposes Salesforce development tools (scan, lint, test, deploy, verify) to AI assistants via the Model Context Protocol, enabling programmatic interaction with Salesforce projects.5 npmMIT