- DATA
- INSIGHT
- DECISION
- SUMMARY
Short answer
How can I validate experiment results with my AI?
Your agent uses the PostHog MCP to pull current experiment data into your chat. You can then ask it to compare variations or summarize progress. This breakdown shows how the data flows from your dashboard to your agent.
The outcomes of your analysis.
Where your data lands.
Using this connector gives you three ways to look at your product tests.
DATA
Active experiment lists
The AI pulls a list of all currently running tests. You get a clear view of what's live in PostHog right now.
INSIGHT
Variation comparisons
Your agent looks at the performance of different groups. It helps you see which version is winning based on the data.
DECISION
Human-led conclusions
The AI flags the trends, but you make the final call on whether to ship a feature. It provides the evidence you need to decide.
SUMMARY
Progress reports
Get a quick text summary of how a specific test is going. You can use these to keep track of long-running experiments.
The workflow
What your AI does when the data arrives.
The AI acts as the bridge between your PostHog dashboard and your chat window.
Find active tests
The agent scans your PostHog project to find which experiments are currently running. This ensures you are looking at live data.
list_experimentsPull experiment details
The AI retrieves the specific data for the tests you identify. It pulls the metrics needed to understand the current state of a test.
list_experimentsCompare performance
Your agent compares the different variations within an experiment. It identifies the differences in user behavior between groups.
list_experimentsSummarize outcomes
The AI turns the raw numbers into a plain English summary. You get a clear picture of whether a test is meeting its goals.
list_experiments
Try it
Copy these to start.
Use these prompts to kick off your analysis.
Starting points
These are opening lines. Swap out the experiment names or project IDs to match your actual PostHog setup.
PostHog Connector
You're all set. Choose your MCP client and follow the setup instructions.
https://edge.vinkius.com/vk_preview_vpYk0NiKfoRYdh0mOuVMNfGBNKFlD4K0TVsVAy1i/mcpClaude Desktop
Follow the steps below to connect in seconds.
- 1In Claude Desktop, open Settings → Connectors.
- 2Click “Add custom connector” and paste the connector link above as the remote MCP server URL.
- 3Click Add and start a new chat — PostHog capabilities are ready to use.
{
"mcpServers": {
"posthog-mcp": {
"url": "https://edge.vinkius.com/vk_preview_vpYk0NiKfoRYdh0mOuVMNfGBNKFlD4K0TVsVAy1i/mcp"
}
}
}List all my active experiments in PostHog.
Show me the results for the 'new_onboarding_flow' experiment.
Which variation is performing better in my current test?
Summarize the progress of the 'checkout_button_color' experiment.
Claude
ChatGPT
Cursor
VS Code
Windsurf
Claude Code
JetBrains
Cline
Start here
Connect PostHog once, then ask.
Provide your credentials once. Vinkius keeps the connection secure and ready for your AI client to use.
Connect PostHog to your AIFAQ
How this task behaves.
- 01
Can the AI change my experiment settings?
No. The AI can only read experiment data. It cannot modify, start, or stop experiments in PostHog.
- 02
Can the AI delete my experiments?
No. The AI has no permission to delete any data or experiments from your PostHog account.
- 03
Can I use this with Cursor?
Yes. Any MCP-compatible client like Cursor, Claude, or Windsurf can use this once you connect it.
- 04
Does the AI see my entire PostHog database?
The AI only accesses the specific experiment data requested through the available tools.
- 05
Can the AI create a new experiment for me?
No. The current tools only allow the AI to list and read existing experiment data.
More questions about PostHog? See everything the PostHog Connector can do
Connect PostHog to Claude, Cursor, ChatGPT & more
