LaunchDarkly MCP server

What the model can do

LaunchDarkly feature flags: projects and environments, flags with their per-environment on/off state, fallthrough and rollouts, rules and prerequisites, flag statuses (new, active, inactive, launched), the audit log of who changed what, and account members; and, when granted, turn a flag on or off in an environment, change its targets, fallthrough or off variation, create, rename, tag, archive or delete flags, request, review and apply approvals, and create, fill or delete segments.

Reads

• launchdarkly_projectsList LaunchDarkly projects with their environments (key, name, colour, critical).
• launchdarkly_environmentsEnvironments of one project: key, name, colour, tags, whether it is marked critical, default TTL, secure mode…
• launchdarkly_flagsFeature flags in a project: key, name, kind (boolean or multivariate), tags, maintainer, variations…
• launchdarkly_flagOne flag in full: its variations, and per environment the on/off state, fallthrough or rollout, off variation, every targeting rule…
• launchdarkly_flag_statusesWhether flags are actually being evaluated: for every flag in one environment its status — new (created under a week ago, never requested), active…
• launchdarkly_audit_logWho changed what, newest first: audit log entries in the words LaunchDarkly uses ("Ariel Flores turned on the flag new-checkout in Production")…
• launchdarkly_membersAccount members: name, email, role (reader, writer, admin, owner or custom), custom roles, teams, MFA, pending invites and when they were last seen.

Acts, when you ask

These change something in LaunchDarkly. The model is told to use them only when you ask for exactly that, and each one is granted by name: per team on a shared connection, by the owner on a personal one.

• launchdarkly_turn_flagTurn a feature flag's targeting on or off in ONE environment.
• launchdarkly_update_flag_targetingChange who a flag serves in ONE environment, one change per call: add or remove individual targets (context keys served one variation)…
• launchdarkly_create_flagCreate a feature flag in a project, in every environment, with targeting off.
• launchdarkly_update_flag_settingsChange a flag's settings across all environments: rename it, change its description, add or remove tags, mark it temporary or permanent…
• launchdarkly_archive_flagArchive a flag (retire it from the flag list and dashboards; SDKs keep evaluating its last configuration)…
• launchdarkly_delete_flagPermanently delete a flag and its rules from every environment of a project.
• launchdarkly_request_flag_changeAsk for approval of a flag change in ONE environment instead of making it, for environments that require approvals: turn on, turn off…
• launchdarkly_review_approvalApprove, decline or comment on a LaunchDarkly approval request, as the token's member.
• launchdarkly_apply_approvalApply an approved LaunchDarkly approval request, making its flag change live in that environment at once.
• launchdarkly_create_segmentCreate an empty standard segment in ONE environment of a project.
• launchdarkly_update_segment_targetsAdd or remove context keys in a segment's included or excluded list, in ONE environment.
• launchdarkly_delete_segmentPermanently delete a segment from ONE environment.

Connect it

First connect LaunchDarkly on Dashboard → Connections, then add Memtro to your client once. The LaunchDarkly tools appear alongside the tools of everything else you have connected; nothing is configured per connector.

claude plugin marketplace add memtro-org/claude-memtro
claude plugin install memtro@memtro --config api_key=sk-memtro-...

{
  "mcpServers": {
    "memtro": {
      "url": "https://platform.memtro.com/api/mcp",
      "headers": {
        "Authorization": "Bearer sk-memtro-..."
      }
    }
  }
}

{
  "mcp": {
    "memtro": {
      "type": "remote",
      "url": "https://platform.memtro.com/api/mcp",
      "headers": {
        "Authorization": "Bearer sk-memtro-..."
      }
    }
  }
}

Who may use it

The key identifies the person, and the server offers that person the tools they may use everywhere else in Memtro. A shared LaunchDarkly connection is used by the teams an administrator granted it to, tool by tool, on the Permissions page; a personal connection is used by its owner alone, and its tools that act stay off until the owner allows them. See Teams and permissions and Connectors.

Ask it like this

• › Is new-checkout on in production, and what percentage of users get it?
• › Who turned off the pricing-v2 flag last Tuesday, and did they leave a comment?
• › Which flags in the web project haven't been requested for a week?
• › List the members with admin or owner roles who don't have MFA on
• › Turn off new-checkout in production, comment: error spike at 14:10

Questions

Is this an official LaunchDarkly MCP server?

No. It is Memtro's MCP server with Memtro's LaunchDarkly connector behind it: the connector reads LaunchDarkly through its API with the credentials you connect on Dashboard → Connections, and LaunchDarkly publishes nothing here. The same endpoint carries every other system you connect and Memtro's memory tools.

What does it cost?

Nothing on top of Memtro. One price per organisation by its number of users, from $50 a month for up to 10 users after a 7-day free trial, with every connector, the MCP server, jobs and memory included. You bring your own model keys.

Can it change anything in LaunchDarkly?

Only when you ask, and only what it has been granted: turn a flag on or off in one environment; change a flag's individual targets, fallthrough (variation or rollout) or off variation in one environment; create a feature flag; rename a flag, change its description, tags, temporary status or maintainer; archive or restore a flag; delete a flag permanently; request approval for a flag change, notifying the reviewers named; approve, decline or comment on an approval request; apply an approved approval request; create a segment in one environment; add or remove context keys in a segment's included or excluded list; delete a segment permanently. Every other tool reads, searches or lists. An administrator grants each acting tool by name per team, and a personal connection acts only after its owner allows it.

Which clients work?

Any MCP client that speaks Streamable HTTP: Claude Code, Cursor, Claude Desktop, OpenCode, Codex, Zed, Windsurf, Cline and the Anthropic and OpenAI SDKs. There is no local process to run; the server is remote and authenticated with your key.

Other MCP servers · Engineering and operations

• GitHub MCP server
• AWS CloudWatch MCP server
• Sentry MCP server
• ServiceNow MCP server
• Okta MCP server
• Twilio MCP server