What can an AI agent do with LaunchDarkly?

A LaunchDarkly AI agent can list and inspect feature flags, check evaluation results for specific users, monitor rollout status across environments, toggle flags on or off, update targeting rules, and create new flags. It's useful for progressive delivery workflows, flag hygiene audits, and incident response. Write actions like toggling flags or changing targeting rules always require your approval.

Will my agent change flags in production without asking?

No. Toggling flags and updating targeting rules are write actions that require your explicit approval in safe and plan permission modes. You can configure auto-approval for trusted low-risk workflows.

Which LaunchDarkly environments can my agent access?

Your agent can access any environment included in your connected LaunchDarkly API key's scope. You can scope tasks to a specific environment (e.g. staging only) in the task instructions.

How is a LaunchDarkly AI agent different from LaunchDarkly's built-in automation?

LaunchDarkly's automation rules handle simple threshold-based triggers within the platform. An AI agent reasons across tools: it reads flag state, correlates with error data from Datadog or PagerDuty, drafts a rollback plan, and posts a summary to Slack - all in one workflow, without you having to stitch services together manually.

Developer tools

LaunchDarkly, run by an agent.

An AI agent that manages, monitors, and automates your feature flag rollouts.

Connect LaunchDarkly Book a demo

What needs your approval

LaunchDarkly never changes anything without you.

Anything that creates, edits or deletes in LaunchDarkly waits for your approval. Approve a recurring task once and it runs unattended after that.

Connecting LaunchDarkly

About 5 minutes

Auth: API key
Setup: About 5 minutes

Where to get access

Create an API access token in LaunchDarkly under Account Settings > Authorization.

app.launchdarkly.com

Example tasks

Real work people hand to an agent that uses LaunchDarkly.

monitor a canary rollout and alert in Slack if error rates spike before expanding to 100%
audit all flags older than 90 days and generate a cleanup report for the team
coordinate an a/b test by updating targeting rules and tracking which variant is live per environment
auto-disable a flag in production when a PagerDuty incident is triggered for the affected service