developer-toolsapi_key

LaunchDarkly

A feature management platform that helps teams build better software faster using feature flags.

Verdict

LaunchDarkly's MCP lets your team manage feature flags, environments, and deployment workflows directly from Switchy. @Mention it to check which flags are live in production, set up automated triggers that toggle flags when conditions change, or audit who has access to which environments. Engineers get instant visibility into flag states without switching tools. Product managers can coordinate releases by creating triggers tied to specific environments. The MCP requires an API key with read and write scopes — you'll need admin access to generate one from LaunchDarkly's account settings.

Common use cases

Audit flag states before a production deploy
Create automated triggers for gradual rollouts
List environments to verify staging configs
Check code references for deprecated flags
Review custom roles during security audits

Integration

Vendor: LaunchDarkly
Category: developer-tools
Auth: API_KEY
Tools: 7
Composio slug: launch_darkly

Tools

Create Flag Trigger Workflow
Tool to create a flag trigger workflow. use when setting up an automated trigger to toggle a flag in a specific environment.
Delete Trigger Workflow
destructive
Tool to delete a specific flag trigger workflow. use after confirming the project key, feature flag key, environment key, and trigger workflow id.
Get Custom Roles
Tool to retrieve a list of all custom roles. use when you need to list roles for permissions audits or configuration reviews. note: results are paginated.
Get Environments
Tool to retrieve a list of all environments within a project. use after confirming the project key; supports pagination.
List Code Reference Repositories
Tool to list code reference repositories. use when retrieving repositories for a specific project; includes pagination.
List Projects
Tool to retrieve a list of all projects. use after confirming credentials to enumerate project keys and metadata.
List Teams
Tool to list all teams in launchdarkly. use after confirming credentials; supports pagination internally.

Setup

Setup guide

11. In Switchy, open your workspace settings and click 'Add Integration' under the MCP section. 2. Select LaunchDarkly from the developer tools category. 3. Go to LaunchDarkly's account settings, navigate to Authorization, and create a new API access token with 'Writer' role (or custom role including projects, environments, flags, and code references read/write). 4. Copy the token and paste it into Switchy's API key field, then click 'Connect'. 5. Switchy will verify the token by listing your projects — if successful, you'll see a green checkmark. 6. Open any Space, type '@LaunchDarkly list projects' and send the message. 7. The MCP should return your project list within seconds, confirming the connection works. 8. You can now invoke any of the seven tools by @mentioning LaunchDarkly and describing what you need in plain language.

What teammates see: by default, memories from LaunchDarkly are scoped to the Space (PROJECT visibility) - you can mark any memory PRIVATE or share it ORG-wide.

Works well with

Top models

Compatibility data appears once enough Spaces have used this MCP together with a given model.

How Switchy teams use it

Not enough Spaces yet to publish anonymised usage stats (we require ≥ 50 Spaces per week).

Starter prompts

List All Projects

@LaunchDarkly list all projects in our account. Show the project key and name for each one.

Open in a Space →

Check Production Environments

@LaunchDarkly get all environments for project key 'mobile-app'. Include environment keys and whether each is a production environment.

Open in a Space →

Create Release Trigger

@LaunchDarkly create a flag trigger workflow for project 'web-platform', flag 'new-checkout-flow', environment 'production'. Trigger should enable the flag when the workflow is invoked.

Open in a Space →

Audit Custom Roles

@LaunchDarkly retrieve all custom roles. Show the role name and a summary of permissions for each.

Open in a Space →

Find Code References

@LaunchDarkly list code reference repositories for project 'backend-services'. Include repository names and the number of flag references in each.

Open in a Space →

Example outputs

Illustrative - representative of the model's voice and quality, not literal recordings.

Prompt

@LaunchDarkly list all feature flags in the production environment for the checkout-service project

Output

This example would retrieve flags from LaunchDarkly's production environment. The response might show: 'Found 12 feature flags in checkout-service/production: express-checkout (ON, targeting 25% of users), apple-pay-button (OFF), one-click-reorder (ON, targeting beta-testers segment), payment-retry-logic (ON for all users)…' along with each flag's current state, targeting rules, and last modified timestamp. The MCP surfaces this data so the AI can reason about deployment state without you opening the LaunchDarkly console.

Notes

Requires read access to the specified project and environment via your API key. LaunchDarkly's API returns paginated results; large flag inventories may need follow-up queries. This example assumes the Get Environments tool is extended or combined with flag-listing logic—confirm your MCP version supports flag enumeration beyond the seven tools listed.

Prompt

@LaunchDarkly create a trigger workflow that turns on the dark-mode flag in staging when a GitHub PR is merged to the feature/ui-refresh branch

Output

This example would configure an automated flag trigger. The response might confirm: 'Created trigger workflow dark-mode-staging-auto in project mobile-app, environment staging. Trigger: GitHub PR merge to feature/ui-refresh. Action: Set dark-mode flag to ON. Workflow ID: 6f8a3c2e. The flag will toggle automatically on the next matching event—no manual dashboard visit required.' The MCP handles the API call to persist this workflow in LaunchDarkly's trigger system.

Notes

Requires write permissions on the target project and environment. Side-effect: the flag will toggle in production traffic once the trigger fires—test in a non-production environment first. LaunchDarkly's trigger workflows depend on external integrations (GitHub, Datadog, etc.) being configured in your account; the MCP cannot set those up for you.

Prompt

@LaunchDarkly show all custom roles and summarize which roles have write access to production environments across our projects

Output

This example would fetch custom roles and analyze permissions. The response might read: 'Retrieved 8 custom roles. Three roles grant production write access: deploy-engineers (all projects), payments-team (checkout-service only), and platform-admins (global). The other five roles are read-only or restricted to staging/dev. Notable: the contractor role was last modified 90 days ago and may need a permissions review.' The AI synthesizes role data into an audit-ready summary without you parsing JSON payloads.

Notes

Requires account-level read access to roles via your API key. Results are paginated; accounts with dozens of roles may need multiple fetches. This example pairs the Get Custom Roles tool with AI reasoning to surface security insights—useful for compliance audits, but the MCP cannot modify role permissions itself.

Use-case deep-dives

Feature flag audit for compliance

When LaunchDarkly MCP wins for quarterly access reviews

A 6-person SaaS team runs quarterly SOC 2 audits and needs to document which engineers can toggle production flags. The LaunchDarkly MCP pulls custom roles and team assignments in one query, letting the security lead cross-reference flag access against the org chart without logging into the vendor UI. This works cleanly if your flag count stays under 200 and you're not running real-time access changes during the audit window. Beyond that threshold, the pagination overhead makes the MCP slower than exporting a CSV from LaunchDarkly's dashboard. If your compliance cadence is quarterly or slower and you want the audit trail inside your AI workspace, this MCP closes the loop in 3 minutes instead of 30.

Incident response flag rollback

When trigger workflows beat manual toggles in production

A 12-person e-commerce platform sees a checkout conversion drop at 2am and suspects a new payment-flow flag. The on-call engineer uses the LaunchDarkly MCP to create a trigger workflow that auto-reverts the flag if error rates spike above 5 percent, then goes back to sleep. This scenario wins when your team already trusts LaunchDarkly's metric integrations and the flag has a clean rollback path. It falls apart if the flag has complex dependencies or if your monitoring lives in a separate tool the MCP can't read. For teams running feature flags in production with metric-based kill switches, this MCP turns a 10-minute scramble into a 90-second safeguard.

Code reference cleanup sprint

When repository scanning finds orphaned flag references

A 4-person mobile team wraps a feature launch and wants to delete stale flags from the codebase before the next sprint. The LaunchDarkly MCP lists code reference repositories and cross-checks them against active flags, surfacing 14 orphaned references the team forgot to remove after the last release. This works if your repos are already connected to LaunchDarkly's code refs feature and you're running the cleanup as a scheduled task, not a real-time gate. If your team doesn't use LaunchDarkly's GitHub integration or your flag lifecycle is under 2 weeks, the manual overhead isn't worth automating. For teams with 20-plus flags and monthly cleanup cycles, this MCP turns technical debt triage into a 5-minute standup agenda item.

Frequently asked

What can the LaunchDarkly MCP do in Switchy?

The LaunchDarkly MCP lets your AI agents manage feature flags, environments, and workflows directly from Switchy. Agents can create or delete flag trigger workflows, audit custom roles and team permissions, list projects and environments, and pull code reference repositories. It's built for teams that want AI to handle LaunchDarkly admin tasks without switching contexts or writing scripts.

Do I need admin access to connect LaunchDarkly?

You need a LaunchDarkly API key with sufficient permissions for the tools you want to use. Creating or deleting trigger workflows requires write access to flags and environments. Listing projects, teams, or custom roles needs read access to those resources. LaunchDarkly's API key scoping is granular, so you can create a key that matches your team's security policy.

Can the MCP toggle feature flags on and off?

No. The MCP creates and deletes trigger workflows that automate flag toggles, but it doesn't directly flip flags. If you need an agent to enable or disable a flag right now, you'll need to use LaunchDarkly's UI or a different integration. The MCP is for managing the automation layer, not runtime flag state.

Why use this instead of LaunchDarkly's web dashboard?

The MCP is faster when you're already working in Switchy and need to audit environments, check which teams have access, or set up a trigger workflow. Instead of context-switching to the LaunchDarkly UI, your agent pulls the data or makes the change inline. It's not a replacement for the dashboard—it's for teams that want AI to handle repetitive LaunchDarkly tasks.

Who on the team should connect the LaunchDarkly MCP?

Whoever manages your LaunchDarkly account and can generate API keys. That's usually a platform engineer or DevOps lead. Once connected, any Switchy workspace member can ask agents to use it, but the API key's permissions control what actions are allowed. If you're worried about accidental changes, scope the key to read-only access first.