Configs Variations

This is a systematic approach to A/B testing your AI configs through LaunchDarkly's MCP server. You clone a baseline variation and override just the model, prompt, or parameters you want to test, keeping everything else identical so you know what actually moved the needle on cost, latency, or quality. The skill enforces the one variable at a time principle and protects your baseline from accidental edits. It's opinionated about cloning over creating from scratch, which makes sense when you're running experiments. The workflow is clear: identify what to optimize, design the test, clone with selective overrides, verify the diff. Best for teams already using LaunchDarkly who want to treat prompt engineering like feature flagging.

skills.sh installs: 1.3K
GitHub stars: 19
last updated: Jul 14, 2026

View on GitHub

Install

$ npx -y skills add launchdarkly/agent-skills --skill configs-variations --agent claude-code

Installs into .claude/skills of the current project.

Files

Config Variations

You're using a skill that will guide you through testing and optimizing configs through variations. Your job is to design experiments, create variations, and systematically find what works best.

Prerequisites

This skill requires the remotely hosted LaunchDarkly MCP server to be configured in your environment.

Primary MCP tool:

create-agent-config with resource: "variation" and sourceVariationKey -- clone a baseline variation with selective overrides (recommended for experimentation)

Alternative MCP tools (for more control):

get-agent-configs -- review existing variations before adding new ones
create-agent-config with resource: "variation" -- create new variations from scratch

Optional MCP tools:

update-agent-config with resource: "variation" -- refine a variation after creation
delete-agent-config with resource: "variation" -- remove variations that didn't work out