Five flag states
off for no one, admins for platform admins, targets for named users and workspaces, percent for a sticky share of users, and on for everyone. Admins see every state except off.
Early access
Features
What runs in production today, and what we're building for early access. Everything marked live already runs Sandboxer and Transformer.
Sticky percentage rollouts
A user's bucket comes from a SHA-256 hash of the flag key and user id, so the same user stays in or out as you raise the percentage from 0 to 100.
User and workspace targeting
Target user ids or verified emails (matched without case) and workspace ids, checked against every workspace a user belongs to. Rules are kept when the state changes.
Kill switch
A single call sets a flag off and keeps its rules and targets, with an optional note saying why.
Audit history
Every change is stored with who made it, when, the state and rules before and after, and a note. The change, its history row and the audit log are written together.
Flags declared in code
Each flag is declared with a key, description, owner and default. New flags start off, and a deploy never overwrites a state someone set.
Fast, safe checks
Checks read an in-memory snapshot refreshed every 5 seconds and right after a change. Unknown, not yet loaded or errored means off.
Private by design
The browser only receives the flags that are on for the caller. The flag list and targeting are never sent to non-admins, and signed-out visitors only see flags that are on for everyone.
CLI, API and agent tools
List, inspect, set, target and kill flags from a CLI or an admin API. In Transformer, admin agents also use MCP tools. Customer API keys never get flag access.
Test helpers
withFlag forces a flag for one test, setFlagOverride for a whole file, and bothFlagStates runs a test with the flag off and then on.
Shareable admin views
The admin list opens a flag in a drawer at its own address, so a reload or a shared link shows the same flag.
Flags tied to metrics
Watch error rates, latency and your own business metrics next to each flag's state and rollout.
Guarded rollouts with auto-rollback
Step a percentage rollout up on a schedule and switch the flag off automatically if a guarded metric regresses.
Segments
Define an audience once, such as beta workspaces or internal staff, and reuse it across flags.
Approvals
Require a second person to approve changes to sensitive flags before they take effect.
SDKs
Server and browser libraries so any app can check Flagger flags with the same off-by-default rules.
Experiments
Split users between variants and compare them on the metrics you pick.
Hosted MCP server
Connect your agent to Flagger directly, with admin-only credentials you can revoke.
Want a say in what comes first?
Tell us how your team ships when you join the waitlist.