LaunchDarkly for AI agents
LaunchDarkly is a feature management platform engineering teams use to control feature flags, rollouts, and experiments in production without redeploying code. Connect it once through Arc0, and your agent, or Claude, ChatGPT and Cursor, can use it through one MCP endpoint, limited to what each user approved.
What agents do in LaunchDarkly.
Check a feature flag's status
Look up a flag's targeting rules and environment status before a release discussion.
Submit a flag change for approval
Create an approval request to change a flag's rollout, since flag changes affect production traffic.
Create a new experiment
Set up a new experiment tied to a feature flag to test a rollout variant.
248 LaunchDarkly actions, graded by risk.
Every LaunchDarkly action is tagged read, write or destructive, so one policy covers the whole app and new actions inherit the right default.
read
125Look things up. Allowed by default.
- launchdarkly.get_ipsGet IPs
- launchdarkly.get_rootGet Root
- launchdarkly.get_tagsGet Tags
- launchdarkly.get_teamGet Team
- launchdarkly.get_tokenGet Access Token
- launchdarkly.get_branchGet Branch
- launchdarkly.get_layersGet Layers
- launchdarkly.get_memberGet Account Member
- launchdarkly.get_metricGet Metric
- launchdarkly.get_tokensGet Tokens
- launchdarkly.list_teamsList Teams
- launchdarkly.get_holdoutGet Holdout
- launchdarkly.get_membersGet Members
- launchdarkly.get_projectGet Project
- launchdarkly.get_segmentGet Segment
- launchdarkly.get_webhookGet Webhook
write
88Create and change things. Allow, or ask the user first.
- launchdarkly.create_teamCreate Team
- launchdarkly.create_layerCreate Layer
- launchdarkly.update_layerUpdate Layer
- launchdarkly.create_metricCreate Metric
- launchdarkly.create_holdoutCreate Holdout
- launchdarkly.create_membersCreate Members
- launchdarkly.create_projectCreate Project
- launchdarkly.create_segmentCreate Segment
- launchdarkly.create_webhookCreate Webhook
- launchdarkly.create_workflowCreate Workflow
- launchdarkly.create_flag_linkCreate Flag Link
- launchdarkly.create_iterationCreate Experiment Iteration
- launchdarkly.update_flag_linkUpdate Flag Link
- launchdarkly.create_experimentCreate Experiment
- launchdarkly.create_extinctionCreate Extinction Events
- launchdarkly.create_repositoryCreate Repository
destructive
35Delete, cancel or archive. Ask first, or deny outright.
- launchdarkly.delete_teamDelete Team
- launchdarkly.delete_tokenDelete Access Token
- launchdarkly.delete_memberDelete Account Member
- launchdarkly.delete_metricDelete Metric
- launchdarkly.delete_projectDelete Project
- launchdarkly.delete_segmentDelete Segment
- launchdarkly.delete_webhookDelete Webhook
- launchdarkly.delete_branchesDelete Branches from Code Reference Repository
- launchdarkly.delete_workflowDelete Workflow
- launchdarkly.delete_flag_linkDelete Flag Link
- launchdarkly.delete_repositoryDelete Repository
- launchdarkly.delete_custom_roleDelete Custom Role
- launchdarkly.delete_destinationDelete Data Export Destination
- launchdarkly.delete_environmentDelete Environment
- launchdarkly.delete_feature_flagDelete Feature Flag
- launchdarkly.delete_metric_groupDelete Metric Group
LaunchDarkly in three steps.
- 01Your users connect LaunchDarklyThey sign in to LaunchDarkly on Arc0 Connect, under your brand, and approve the access you ask for.
- 02You set the rulesReads run, writes like “create Team” can wait for the user, and “delete Team” can be denied outright.
- 03Any agent can actYour agent calls LaunchDarkly through the Arc0 SDK or MCP, and so can Claude, ChatGPT and Cursor. Every call lands on the audit log.
await arc0.policies.set('launchdarkly', { read: 'allow', write: 'ask', // create_team destructive: 'deny', // delete_team }) # Claude Code: the same connection, one URL $ claude mcp add --transport http arc0 \ https://mcp.arc0.ai/u/u_8f2
How LaunchDarkly connects.
Users sign in to LaunchDarkly on Arc0 Connect and approve the scopes you request. Build with Arc0’s LaunchDarkly OAuth app, or bring your own so the LaunchDarkly consent screen names you. Tokens refresh automatically, and you can export them whenever you want.
The same LaunchDarkly connection serves your agent over MCP and your own backend over REST and the proxy, so a user connects once. How Arc0 handles credentials →
- AUTH
- API key · OAuth 2.0
- CREDENTIALS
- Per-tenant encrypted vault
- MODEL SEES
- Results only, never credentials
- AUDIT LOG
- Every call, on every plan
Use LaunchDarkly from any agent.
LaunchDarkly and Arc0, answered.
Can I use LaunchDarkly with Claude, ChatGPT or Cursor?
Yes. Connect LaunchDarkly to Arc0 once, then add your Arc0 MCP URL to Claude, ChatGPT, Cursor, Claude Code or any other remote-MCP client. Each assistant only gets the LaunchDarkly actions you allow.
How do users connect LaunchDarkly?
Users sign in to LaunchDarkly on Arc0 Connect and approve the scopes you request. Build with Arc0’s LaunchDarkly OAuth app, or bring your own so the LaunchDarkly consent screen names you. Tokens refresh automatically, and you can export them whenever you want.
Which LaunchDarkly actions can my agent take?
248 in total: 125 read, 88 write and 35 destructive, such as “create Team”. Your policies decide which of them each agent may call.
Can I stop my agent from deleting things in LaunchDarkly?
Yes. Actions like “delete Team” are graded destructive. Set destructive actions to deny, or to ask so the user approves each one, and blocked calls still show up on the audit log.
Can my own backend call LaunchDarkly too?
Yes. The same LaunchDarkly connection is available over REST and through the Arc0 proxy, so your product and your agent share one connection per user.
Plug LaunchDarkly into your agent.
Your users connect LaunchDarkly once, under your brand. Your agent gets 248 actions behind your policies, with every call on the record.
Free to build · MCP + REST · Audit log on every plan