LaunchDarkly Integration for AWS AI-DLC Workshop

Add feature management to the AWS AI-DLC Workshop using Agent Skills.

---

Workshop Flow

Complete these tutorials in order:

When	Tutorial	What You'll Do
After starting Inception	LaunchDarkly Setup	Create account, get API token (2 min)
During Code Generation	Construction Phase	Ask Kiro to add AI recommendations feature
After Build and Test	Construction Phase	Connect LaunchDarkly, control the AI model
After Operations Overview	Operations Phase	Add feature flags to control issues

Inception Phase
    │
    └── ★ LaunchDarkly Setup (just get your API token)
    │
Construction Phase (after Build and Test)
    │
    └── ★ Kiro CLI does the work:
    │       /mcp add launchdarkly ...
    │       "Set up LaunchDarkly for this workshop..."
    │       → Creates project, AI Config, stores SDK key
    │
Operations Phase (after Overview)
    │
    └── ★ Feature Flags for issue control

---

Agent Skills

This workshop uses Agent Skills - text-based playbooks that teach Kiro how to manage LaunchDarkly.

Included Kiro Steering

The .kiro/steering/launchdarkly-ai-configs.md file teaches Kiro the LaunchDarkly workflows, referencing skills from the Agent Skills repo.

Available Skills

Skill	What It Does
/aiconfig-projects	Create a LaunchDarkly project for the workshop
/aiconfig-create	Create AI Configs with model + instructions
/aiconfig-variations	Add variations for A/B testing
/aiconfig-targeting	Change which users get which variation
/aiconfig-tools	Attach function-calling tools

---

Quick Reference

Your Keys

Key	Starts With	Where to Get It	What It's For
API Token	api-	Account Settings → Authorization	Agent Skills / MCP
SDK Key	sdk-	Project Settings → Environments → Test	Your application code

Set API Token

export LAUNCHDARKLY_ACCESS_TOKEN="api-YOUR-TOKEN"

Store SDK Key in AWS

aws ssm put-parameter \
  --name "/icode/launchdarkly/sdk-key" \
  --value "sdk-YOUR-KEY" \
  --type SecureString

---

What You'll Build

Construction Phase: AI Configs

Switch AI models using Agent Skills:

You (in Kiro)
    │
    │  /aiconfig-targeting
    │  "Change default for aidlc-agent to opus"
    ▼
LaunchDarkly (updates targeting)
    │
    ▼
Your Agent (automatically uses new model)
    │
    │  No code change needed!
    ▼
AWS Bedrock (runs Claude Opus instead of Sonnet)

Operations Phase: Feature Flags

Control the three workshop issues:

Flag	Controls
catalog-service-enabled	Catalog service replicas (0 or 1)
enable-security-group-fix	Security group rule (missing or present)
use-correct-health-check	ALB health path (/health or /actuator/health)

---

Files in This Repo

.kiro/
└── steering/
    └── launchdarkly-ai-configs.md   # Kiro steering for Agent Skills

workshop-docs/
├── launchdarkly-setup.md            # Step 1: Account, keys, setup
├── construction-launchdarkly.md     # Step 2: AI Configs for model switching
└── operations-launchdarkly.md       # Step 3: Feature flags for issue control

examples/
├── aiconfig-agent.py                # Python example with Bedrock
└── frontend-flags.tsx               # React example with feature flags

---

Links

• AWS AI-DLC Workshop
• LaunchDarkly Agent Skills
• LaunchDarkly Free Trial
• LaunchDarkly Docs