RudderStack Transformation

@RudderStack Transformation list all published libraries in our workspace and show me which ones are used for user enrichment

Found 8 published libraries in your workspace. Three are tagged for user enrichment: 'geo-lookup' (v2.1, last updated 14 days ago), 'segment-classifier' (v1.3, updated 6 days ago), and 'email-validator' (v3.0, updated yesterday). The geo-lookup library is currently imported by 4 active transformations. The other enrichment-related libraries include 'currency-converter' and 'timezone-mapper', which might also be relevant depending on your use case.

This example demonstrates the MCP's read capabilities for workspace inventory. Requires OAuth2 with read:library scope. The response reflects your workspace's actual library state — if you have no published libraries, the list will be empty. Useful for auditing code reuse before creating new transformations.

@RudderStack Transformation create a new transformation called 'filter-test-events' that drops all events where context.testMode is true, but don't publish it yet

Created unpublished transformation 'filter-test-events' (ID: tr_8x4k2m9p). The transformation includes a JavaScript function that checks `event.context.testMode` and returns null to drop matching events. Since publish=false, this transformation is saved as a draft and won't process any live event traffic. You can test it in the RudderStack dashboard, then use the publish action when ready to activate it on your event pipelines.

This write example shows how to safely create transformations without immediately affecting production traffic. The default publish=false behavior is a safety feature — you must explicitly publish to route events through the transformation. Requires write:transformation scope. Once published, the transformation becomes immutable; edits create new versions.

@RudderStack Transformation get transformation tr_5n8j3k1m and analyze its code to suggest performance improvements for high-volume event streams

Retrieved transformation 'enrich-user-profiles' (v4, published 3 weeks ago). The code performs three external API calls per event: Clearbit lookup, IP geolocation, and company enrichment. For high-volume streams, this creates a bottleneck. Recommendations: (1) Move the Clearbit lookup to a library with caching to avoid redundant calls for known emails. (2) Batch the geolocation requests or use a local MaxMind database. (3) Consider async processing for company enrichment if real-time isn't required. The transformation currently has no error handling for API timeouts.

This synthesis example pairs the MCP's retrieval capability with AI reasoning to provide actionable insights. The AI can read transformation code and suggest optimizations, but cannot directly measure runtime performance — you'll need RudderStack's observability tools to validate improvements. Useful for code reviews before scaling event volumes.

When RudderStack Transformation MCP speeds up data QA

A 6-person growth team at a B2B SaaS company ships 3-5 event schema changes per sprint. Their data analyst needs to verify that transformations applied to Segment-style events are working before they hit the warehouse. This MCP is the right call: the analyst can query transformation code directly in Switchy, compare versions when a metric breaks, and pull library dependencies without opening the RudderStack UI. The 12-tool scope covers the full read-write loop—create, publish, fetch by ID, list libraries. The OAuth2 flow means the analyst authenticates once and the whole team inherits access. If your team ships fewer than 2 schema changes per month, the setup overhead outweighs the benefit. For teams iterating on event pipelines weekly, this MCP turns transformation audits from a 20-minute context-switch into a 2-minute Switchy query.

Using this MCP to audit PII handling in transformations

A 3-person data governance team at a fintech startup needs to prove to auditors that no PII leaks into third-party destinations. They run quarterly reviews of every RudderStack transformation that touches user events. This MCP fits: the compliance lead can list all published transformations, retrieve each one's code and version history, and cross-reference libraries that might contain redaction logic. The Get Transformation Version tool is critical here—auditors want to see what was live on a specific date. The MCP doesn't parse the JavaScript or Python for you, but it surfaces the code in Switchy where the team can grep for field names or run static analysis. If your compliance process is annual or ad-hoc, exporting CSVs from RudderStack is simpler. For quarterly or monthly reviews, this MCP cuts the audit prep from a full day to under an hour.

When this MCP helps new hires learn transformation patterns

A 10-person engineering team at a marketplace startup hires 2 junior backend engineers who need to understand how events are enriched before they reach analytics tools. The tech lead wants them to read existing transformations and libraries without giving them write access to production RudderStack. This MCP is borderline useful: the OAuth2 scope is workspace-wide, so you can't restrict juniors to read-only in Switchy unless RudderStack's own RBAC supports it. If it does, the MCP shines—juniors can search transformations by keyword, pull library code to see reusable patterns, and ask Switchy to explain what a transformation does. The List All Libraries and Get Library by ID tools make it easy to trace dependencies. If your RudderStack workspace lacks granular permissions, this MCP won't enforce read-only access. For teams with RBAC configured, it's a low-friction way to onboard engineers to your event pipeline in under a week.