> ## Documentation Index > Fetch the complete documentation index at: https://docs2.growthbook.io/llms.txt > Use this file to discover all available pages before exploring further. # Feature flag skills > The GrowthBook Agent Skills for feature flags: create, target, roll out, ramp, monitor, and clean up flags from your AI agent, with draft → review → publish built in. These [Agent Skills](/integrations/ai-agents/agent-skills) cover the full feature flag lifecycle from your AI agent. Install the plugin and run [`/growthbook:gb-setup`](/integrations/ai-agents/agent-skills#installation) first. Every flag change in GrowthBook goes through a **draft revision** before it goes live. The skills below stage their changes on a draft and hand off to `flag-publish` to ship it: the same draft → review → publish flow you'd use in the app. See [Publishing & approval flows](/features/publishing-and-approval-flows) for the concepts. The example prompts can be typed verbatim, or your agent may run these skills automatically when it detects matching intent. ## Revision lifecycle These three skills manage the draft → review → publish flow that every other flag change flows through. | Skill | What it does | Example prompt | | ---------------- | -------------------------------------------------------------------------------------- | ------------------------------------------------------- | | `flag-revisions` | List and inspect open drafts, see who owns them and their approval status. | "What drafts are open on the `checkout-redesign` flag?" | | `flag-review` | Request a review on a draft, or approve / request changes / comment as a reviewer. | "Request review on my draft for `new-pricing`" | | `flag-publish` | Publish a draft live, resolve merge conflicts, discard, or revert to a prior revision. | "Publish the approved draft for `new-pricing`" | ## Operations Create flags and manage their metadata, default value, environment state, and lifecycle. | Skill | What it does | Example prompt | | -------------------- | ------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------- | | `flag-create` | Create a new flag: collision check, value type, environments, default value. Ships disabled everywhere. | "Create a boolean flag `new-checkout-flow` defaulting to false" | | `flag-metadata` | Update a flag's description, owner, project, tags, custom fields, or JSON schema. | "Tag `dark-mode` as `ui` and set me as the owner" | | `flag-default-value` | Change the fallback value served when no targeting rules match. | "Change the default value of `max-upload-mb` to 50" | | `flag-toggle` | Enable or disable a flag in a specific environment. | "Turn off `new-checkout-flow` in production" | | `flag-prerequisites` | Gate an entire flag on another boolean flag being on. | "Only evaluate `beta-dashboard` when `beta-program` is on" | | `flag-cleanup` | Archive or delete a stale flag, walking through code-site inlining first. | "Clean up the `old-banner` flag" | Related concepts: [Feature flag basics](/features/basics), [Stale flag detection](/features/stale-detection), [Code references](/features/code-references). ## Rules Control who gets which value, and when. For a new A/B test, see also the [Experiment skills](/integrations/ai-agents/agent-skills/experiments). | Skill | What it does | Example prompt | | ----------------- | --------------------------------------------------------------------------------------------- | ---------------------------------------------------------------- | | `flag-rules` | List rules in evaluation order, delete a rule, reorder, or route to the right rule skill. | "Show me the rules on `new-checkout-flow`" | | `flag-targeting` | Add, edit, or remove force and percentage-rollout rules, with conditions and saved groups. | "Roll out `new-checkout-flow` to 10% of US users" | | `flag-schedule` | Time-gate a rule with a start and/or end datetime for automatic activation. | "Enable the sale banner Friday at 9am and turn it off Monday" | | `flag-ramp` | Multi-step ramp schedule that increases coverage over time, with per-step intervals or gates. | "Ramp `new-checkout-flow` from 10% to 100% over a week" | | `flag-monitoring` | Monitored "safe rollout": a ramp with guardrail metrics and optional auto-rollback. | "Roll out `new-checkout-flow` safely and watch checkout latency" | | `flag-experiment` | Add an experiment-ref rule to run an A/B test through the flag. | "Run an experiment through `new-checkout-flow`" | Related concepts: [Targeting & rules](/features/rules), [Targeting conditions](/features/targeting), [Safe rollouts](/features/safe-rollouts), [Environments](/features/environments). ## Discovery Find and audit flags, and trace how they depend on one another. Both skills are read-only. | Skill | What it does | Example prompt | | ------------- | -------------------------------------------------------------------------------------- | ------------------------------------------------------- | | `flag-search` | Search, list, and audit flags by project, tag, owner, environment state, or staleness. | "List all stale flags owned by me" | | `flag-graph` | Trace a flag's prerequisites, its dependents, and any linked experiments and holdouts. | "What depends on `beta-program`? Is it safe to delete?" | ## Next steps * [Experiment skills](/integrations/ai-agents/agent-skills/experiments): design, launch, analyze, and stop tests * [Agent Skills overview](/integrations/ai-agents/agent-skills): install and setup * [Feature flag documentation](/features/basics): the underlying concepts [Join our community Slack for additional tips, help, and announcements](https://join.slack.com/t/growthbookusers/shared_invite/zt-2xw8fu279-Y~hwnfCEf7WrEI9qScHURQ)