Bring your AI agents into your release process
Now your agents can create flags, run experiments, and ship rollouts inside the same controls your team already uses. Works with Claude Code, Cursor, Codex, or any agent that speaks MCP or HTTP.
Trusted by 3,000+ companies worldwide
Connect any agent to GrowthBook in minutes
Get a GrowthBook API Key
Sign up for a free account or connect to your self-hosted instance. Generate an API key from account settings.
Connect your agent
MCP server: paste the JSON block into your AI tool’s MCP settings.
REST API: Hit the API with HTTP actions
Prompt or script
Ask your AI assistant to create flags, check experiments, analyze past experiments. No switching tabs.
Use your agents from idea to clean-up
Create feature flags and experiments
Configure rollouts
Query your analytics
Decide and ship winners
Clean up stale code
Ship with AI speed your team can trust
Your team and agents, in one place
Agent drafts land in your team's review queue. Team rollouts show up in agent status checks. A single backend and source of truth means no reconciliation or drift.
The same guardrails for humans and agents
Every agent action goes through the same approval workflows, role-based permissions, environment scoping, and audit logs your team already uses. No need to maintain a second guardrail system.
The full lifecycle, from wherever you work
Create flags, configure ramps, run experiments, query analytics, conclude winners, and clean up stale code. Every stage works from the GrowthBook app, from your editor over MCP, or from any HTTP client over REST.
One warehouse-native platform for modern product development
Experimentation
Feature Flags
Product Analytics
“GrowthBook gave us a modern experimentation and release platform that actually fits how Dropbox works. We can run analytics directly on our data lake, roll features out safely in stages, and support teams across different stacks without duplicating data or tooling.”
“Being able to turn a feature on and off with a flip of a switch is fantastic... That’s so much easier than having to do a deploy or a roll-back.”
“People only see the wins, but there’s actually greater value in avoiding losses. We’ve stopped changes that could have cost millions.”
“GrowthBook has changed the way we think about experiments... It allowed us to uplevel our code, speed up decision-making, and focus on what we do best.”
“Experimentation showed what customers actually do rather than what we assume they’ll do.”
“We don’t need any code changes, we don’t need an app release. We just configure the new tests and launch right away.”
"We are always experimenting now. It’s a natural part of product development. This is due to GrowthBook and the ease of usage both in the UX and in the seamless integration with Snowflake/DWH."
"GrowthBook's results speak for themselves. Every time we do a test, we see benefits for our audiences and our partners. These posters are our one shot, and we wouldn't want to fly blind."
“GrowthBook lets us build experiments exactly how we want. The ability to target based on culture and geography, as granular as needed, is a major win for us.”
"GrowthBook's results speak for themselves. Every time we do a test, we see benefits for our audiences and our partners. These posters are our one shot, and we wouldn't want to fly blind."
"We are always experimenting now. It’s a natural part of product development. This is due to GrowthBook and the ease of usage both in the UX and in the seamless integration with Snowflake/DWH."
“GrowthBook gave us a modern experimentation and release platform that actually fits how Dropbox works. We can run analytics directly on our data lake, roll features out safely in stages, and support teams across different stacks without duplicating data or tooling.”
FAQs
GrowthBook is built agent-ready from the API up. Anything your team can do in the GrowthBook app, your agents can do through the MCP server or REST API. That includes creating and updating feature flags, managing revisions and approvals, configuring ramp schedules and targeting, setting up experiments from your org's templates, monitoring running experiments through the Decision Framework, querying product metrics and dashboards through MCP or REST, concluding experiments and shipping winners, and finding and archiving stale flags.
An agent-native platform exposes the same surface to agents that it exposes to humans, with the same permissions and audit trail behind both. GrowthBook does that through an open MCP server and REST API that connect to the same backend as the visual app. Your agents are not a side channel. They are first-class users of the platform.
Any AI tool that supports the Model Context Protocol works with the GrowthBook MCP server, including Claude Code, Cursor, VS Code, and Codex. For tools without MCP support, custom agents you build yourself, or any HTTP client, the REST API exposes the same functionality.
MCP (Model Context Protocol) is a standard that connects AI models with external software, databases, workflows, and tools. GrowthBook launched the first open-source production MCP server for experimentation and product development in early 2025. It brings feature flags and testing into AI development environments so engineering teams can measure what they ship without breaking flow.
The MCP server works with agentic clients that support the Model Context Protocol, like Claude Code, Cursor, VS Code, and Codex. The REST API works with any programmatic method, including scripts, CI jobs, internal tools, and custom agents you build yourself. Both hit the same backend with the same permissions, approval flows, and audit trail. Most teams use both: the MCP server lives in their editor, and the REST API powers their automation.
Yes, both the MCP server and the REST API work with self-hosted GrowthBook. Point your API key at your instance URL instead of api.growthbook.io, and the same code that runs in the cloud runs on your hardware. See the self-hosting guide.
GrowthBook applies the same controls to AI agents that it applies to your team: role-based permissions, scoped API keys, approval workflows, environment limits, and audit logs. Every agent action through GrowthBook's MCP server or REST API goes through these guardrails automatically, with no extra configuration.
GrowthBook takes privacy and security seriously. The GrowthBook MCP server runs locally on your machine. It connects only to your GrowthBook instance using your API key, with no telemetry and no third-party calls. The REST API works the same way. Your prompts and source code stay between you and the AI tool you connect, so review that tool's privacy policy separately.
Yes. A GrowthBook instance, cloud or self-hosted, is required before you connec either the MCP server or the REST API. Sign up free, then visit our MCP server docs for tool-specific setup instructions and API docs for the REST reference.
Ready to ship faster?
No credit card required. Start with feature flags, experimentation, and product analytics — free.
