LaunchDarkly

LaunchDarkly

LaunchDarkly is a category anchor for governed runtime control: its first-party MCP can inspect and change feature flags and AgentControl configurations by project and environment, while also querying observability and metrics surfaces.

Analytics, SEO & experimentationSystem of intelligenceExtensible SaaS
An illustration of access surfaces surrounding a governed system.
System roleSystem of intelligence
Access maturityExtensible SaaS
MCP supportOfficial MCP server
Reference updatedNot stated by publisher
BYO-UI reviewedJul 22, 2026

Concrete capability record

Data, retrieval, actions, identity, and operating limits

A field-by-field summary of what the reviewed first-party references actually support. Publisher update dates and BYO-UI review dates are shown separately below.

01 · Records and state owned

Projects and environments; feature flags, variations, and targeting; AgentControl configurations; observability logs, traces, errors, and dashboards; and supported metrics tools.

02 · Data you can retrieve

Inspect flags, targeting, and AgentControl state for permitted projects and environments; query supported logs, traces, errors, dashboards, and metrics.

03 · Actions and write paths

Create and modify flags and AgentControl configurations, including environment targeting changes, and delete those objects when the authenticated role permits. Local MCP can be narrowed to read scope or an explicit tool allowlist.

04 · Authentication and permissions

Hosted MCP uses LaunchDarkly OAuth. Local MCP uses an API access token whose assigned role determines project/environment reach; LaunchDarkly recommends Writer, Developer, or a custom role for full flag and AgentControl management, and a Reader role plus `--scope read` for read-only use.

05 · Monitoring, approval, and recovery

Environment approvals can govern targeting and other covered changes on select plans, but they do not trigger for creating, archiving, or deleting flags or AgentControl configs, selected metadata edits, experiment creation/deletion, or segment creation/deletion. Custom roles must deny those uncovered actions when required.

06 · Limits and caveats

Hosted MCP is unavailable in federal and EU instances, which require the local server. Local deployment needs ECMAScript 2020+, stores responsibility for token handling with the operator, and only achieves read-only behavior when both role scope and server scope are constrained. Required approvals are available only on select plans.

Agent access

MCP support

Official MCP serverLaunchDarkly-hosted OAuth MCP plus local server optionsCurrent documented service

Support: Official MCP server

Read scope: Authorized feature flags, targeting, AgentControl configurations, observability logs/traces/errors/dashboards, and MCP-exposed metrics across selected projects and environments.

Write scope: Authorized creation, update, targeting, archival, and deletion operations for flags and AgentControl configurations; observability access remains query-oriented. Local sessions can expose only selected tools.

Authentication: OAuth for the hosted MCP server; scoped LaunchDarkly API token for the local server.

Approval boundary: Approval requirements are configured per environment and can require reviewer actions, counts, and consensus, but they do not cover object creation, archival, deletion, or specified metadata changes. BYO-UI should add explicit confirmation for those uncovered operations and for any permanent flag deletion.

Protocol does not erase product boundaries.

Confirm the current tool catalog, plan, region, scopes, rate limits, terms, and write behavior before implementation.

Editorial assessment

Access-maturity dimensions

A comparative architecture lens—not a quality score, market ranking, or buying recommendation. Scale: 1–5.

Data access4
Action access4
Event access4
Identity4
Governance4
Agent access4
UI extensibility4
Portability3
Observability3
Documentation4

Proposed customer-shaped experiences

What customers could create on top.

The output could be an export, report, graph, artifact, application, agent, workflow, or downstream feed. These proposals are derived from documented access—not claims that LaunchDarkly ships them.

Single-platform patterns

  • Experiment and release decision workbench
  • Executive insight brief
  • Experiment review board
  • Anomaly investigation room
Use in the brief builder →

Multi-platform compositions

  • LaunchDarkly + warehouse or semantic layer: executive insight brief
  • LaunchDarkly + CRM or lifecycle platform + work-management system: cross-system decision workspace
Explore composition recipes →

Evidence and dates

First-party references, with publisher and review dates separated

“Publisher updated” is shown only when the page exposes an update date. “BYO-UI reviewed” records when this research checked the reference. A missing publisher date is reported as missing—not replaced with the review date.

9 recorded sources
Platform core validationRepresentative source
https://launchdarkly.com/docs/home/getting-started/mcp ↗

Hosted and local deployment; feature, AgentControl, observability, and metrics scope; environment boundaries; approval exceptions; irreversible deletion risk

Publisher updated
Not stated by publisher
BYO-UI reviewed
Jul 22, 2026

Full first-party MCP deployment and governance review completed 2026-07-22

APISource inventory
https://launchdarkly.com/docs/api ↗

First-party REST API reference.

Publisher updated
Not stated by publisher
BYO-UI reviewed
Jul 10, 2026

Inventory source: structurally normalized; content was not individually reopened in this pass.

DocumentationSource inventory
https://launchdarkly.com/docs ↗

First-party platform documentation.

Publisher updated
Not stated by publisher
BYO-UI reviewed
Jul 10, 2026

Inventory source: structurally normalized; content was not individually reopened in this pass.

HomepageSource inventory
https://launchdarkly.com/ ↗

Product overview.

Publisher updated
Not stated by publisher
BYO-UI reviewed
Jul 10, 2026

Inventory source: structurally normalized; content was not individually reopened in this pass.

MCP / agent-role validationTargeted review
https://launchdarkly.com/docs/home/getting-started/mcp-hosted ↗

Official documentation confirms feature-flag, AgentControl, and observability tools; local deployment supports additional environments.

Publisher updated
Not stated by publisher
BYO-UI reviewed
Jul 10, 2026

Targeted first-party MCP/agent-role claim review; exact live reachability was not independently retested for every URL.

MCP deployment guideSource inventory
https://launchdarkly.com/docs/home/getting-started/mcp-local ↗

Local deployment, authentication, permission examples, and exposure restrictions

Publisher updated
Not stated by publisher
BYO-UI reviewed
Jul 22, 2026

Local deployment, authentication, permission examples, and exposure restrictions reviewed from live first-party material on 2026-07-22.

Governance docsSource inventory
https://launchdarkly.com/docs/home/releases/approvals-settings ↗

Approval configuration, required role actions, and enforcement exceptions

Publisher updated
Not stated by publisher
BYO-UI reviewed
Jul 22, 2026

Approval configuration, required role actions, and enforcement exceptions reviewed from live first-party material on 2026-07-22.

Lifecycle docsSource inventory
https://launchdarkly.com/docs/home/flags/delete ↗

Permanent flag deletion lifecycle and runtime consequence

Publisher updated
Not stated by publisher
BYO-UI reviewed
Jul 22, 2026

Permanent flag deletion lifecycle and runtime consequence reviewed from live first-party material on 2026-07-22.

Verified fact

Tied to cited first-party evidence reviewed for this profile.

Source inventory

Official links recorded for deeper research but not necessarily reopened endpoint by endpoint.

Editorial assessment

System role, maturity interpretation, and architectural boundary.

Proposed design

Interface patterns and compositions—not vendor product claims.