feat(mcp): Track report_user_action from MCP separately if it's from the wizard

[robbie-c]

See PostHog/wizard#291

Problem

We need to track which client is making requests to the MCP server (e.g., PostHog Wizard) for better source attribution and analytics. Currently, we extract the client identifier from the User-Agent header but don't forward it in outgoing API requests.

Changes

• Extract posthog/foo identifier from incoming User-Agent headers in the MCP request handler
• Store the client identifier in request context properties
• Pass client identifier through to the ApiClient configuration
• Update User-Agent headers in outgoing API requests to include the client identifier (e.g., posthog/mcp-server; for posthog/wizard)
• Add WIZARD event source to the Python event tracking system to properly categorize requests from PostHog Wizard
• Update get_event_source() to detect and classify wizard requests based on User-Agent

How did you test this code?

The changes are straightforward data flow modifications:

• Client identifier extraction uses standard regex matching on User-Agent headers
• User-Agent header construction is a simple string concatenation
• Event source detection follows the existing pattern used for terraform-provider detection
• No new external dependencies or complex logic introduced

Existing tests should continue to pass as this is additive functionality that doesn't break existing behavior.

Publish to changelog?

No

https://claude.ai/code/session_01AJEWnwWhZPuugRrFz9MgQ1

[greptile-apps]

_{6 files reviewed, no comments}

_{Edit Code Review Agent Settings | Greptile}

[Copilot]

Pull request overview

This PR implements client identifier tracking to enable source attribution for requests from different MCP clients (e.g., PostHog Wizard). The system extracts client identifiers from incoming User-Agent headers and forwards them in outgoing API requests.

Changes:

• Added client identifier extraction from User-Agent headers and propagation through request context
• Updated User-Agent headers in outgoing API requests to include the originating client identifier
• Added WIZARD event source classification in the Python event tracking system

Reviewed changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
services/mcp/src/mcp.ts	Added clientIdentifier to RequestProperties type and ApiClient configuration
services/mcp/src/lib/constants.ts	Created getUserAgent() function to format User-Agent with optional client identifier
services/mcp/src/index.ts	Implemented regex extraction of client identifier from incoming User-Agent header
services/mcp/src/api/fetcher.ts	Updated to use getUserAgent() function with client identifier in API requests
services/mcp/src/api/client.ts	Updated ApiConfig type and User-Agent header construction to support client identifier
posthog/event_usage.py	Added WIZARD event source enum value and detection logic in get_event_source()

Comments suppressed due to low confidence (2)

services/mcp/src/index.ts:208

• The regex pattern /posthog\/(\S+)/ will match and capture everything after 'posthog/' until the first whitespace, which could include unwanted characters like semicolons, commas, or version numbers. For example, 'posthog/wizard;version:1.0' would capture 'wizard;version:1.0'. Consider using /posthog\/([a-z0-9-]+)/i to capture only alphanumeric characters and hyphens, which is more appropriate for identifier matching.

    const clientIdentifierMatch = clientUserAgent.match(/posthog\/(\S+)/)

services/mcp/src/index.ts:209

• Using clientIdentifierMatch[0] returns the entire matched string (e.g., 'posthog/wizard'), but you're capturing a group with parentheses. If you want just the identifier part without the 'posthog/' prefix, use clientIdentifierMatch[1] to access the first capture group. If you want to keep 'posthog/' in the identifier, the capturing group is unnecessary.

    const clientIdentifierMatch = clientUserAgent.match(/posthog\/(\S+)/)
    const clientIdentifier = clientIdentifierMatch ? clientIdentifierMatch[0] : undefined

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[github-actions]

🎭 Playwright report · View test results →

⚠️ 1 flaky test:

• Can submit the signup form multiple times if there is a generic email set (chromium)

These issues are not necessarily caused by your changes.
Annoyed by this comment? Help fix flakies and failures and it'll disappear!