- STATUS
- DATA
- SCOPE
- HUMAN
Short answer
How can I audit my feature flags using AI?
Your agent pulls live metadata and environment lists from GrowthBook to check deployment status and flag rules. You get a clear view of your current configurations through simple text queries. This breakdown shows you how to verify your setup.
Audit outcomes.
What you'll find.
The specific data your agent retrieves during an audit.
STATUS
Deployment visibility
The AI checks which environments have specific features enabled. You'll know exactly what is live in production versus staging.
DATA
Configuration details
Your agent pulls specific metadata for any flag. You get the exact rules and values currently in effect.
SCOPE
Project overview
The AI lists all available environments and features within a project. This helps you map out your entire experimentation setup.
HUMAN
Manual adjustments
The AI only reads data. You still handle the actual toggling or changing of feature values in the GrowthBook UI.
The workflow
What your AI does when the request arrives.
The agent translates your questions into specific data requests.
Identify the project
The agent finds the correct project context to ensure it's looking at the right feature set.
get_projectMap the environment
It pulls a list of all available environments to confirm where your flags are deployed.
list_environmentsLocate the feature
The agent scans your feature list to find the specific flag you're asking about.
list_featuresExtract settings
It grabs the specific configuration and metadata for the target feature.
get_feature
Try it
Copy these to start.
Use these prompts to begin your audit.
Starting points
Swap out the project names and feature flag keys with your actual GrowthBook identifiers.
GrowthBook Connector
You're all set. Choose your MCP client and follow the setup instructions.
https://edge.vinkius.com/vk_preview_KNp7r3vX1MWcig4imXo7kmsnq96l2Q8qFEcS5YIM/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 — GrowthBook capabilities are ready to use.
{
"mcpServers": {
"growthbook-mcp": {
"url": "https://edge.vinkius.com/vk_preview_KNp7r3vX1MWcig4imXo7kmsnq96l2Q8qFEcS5YIM/mcp"
}
}
}List all environments in my Production project.
What are the current settings for the 'new_checkout_flow' feature?
Show me all features currently active in the Staging environment.
Get the metadata for the 'search_algorithm_v2' flag in my main project.
Claude
ChatGPT
Cursor
VS Code
Windsurf
Claude Code
JetBrains
Cline
Start here
Connect GrowthBook once, then ask.
One connection gives your agent access to your feature data. Your credentials stay encrypted, and you can start auditing immediately.
Connect GrowthBook to your AIFAQ
How the audit task behaves.
- 01
Can the AI change my feature flag values?
No. The agent can only read metadata and environment lists. You must use the GrowthBook dashboard to change any settings.
- 02
Can I use this to create new features?
No. The current tools only allow for retrieving existing project, environment, and feature data.
- 03
Does the AI see all my projects?
The AI can see any project it has permission to access through the provided credentials.
- 04
Can the AI delete an environment?
No. There are no tools available to perform deletions or modifications to your GrowthBook setup.
- 05
How does the AI know which environment to check?
You specify the environment name in your prompt, and the agent uses the environment list to find the match.
More questions about GrowthBook? The Connector page answers them. See everything the GrowthBook Connector can do
Connect GrowthBook to Claude, Cursor, ChatGPT & more
