The Agent Skills Directory

[SAFE]: The skill is strictly instructional, providing a five-stage framework for refactoring code to use managed AI configurations. It does not perform any automated or hidden operations.
[EXTERNAL_DOWNLOADS]: The skill references official LaunchDarkly SDKs and provider-specific helpers (e.g., launchdarkly-server-sdk-ai, @launchdarkly/server-sdk-ai-openai). These are well-known, trusted vendor resources.
[CREDENTIALS_UNSAFE]: The skill handles authentication securely by instructing the use of environment variables (LD_SDK_KEY) and provides safe initialization patterns for offline or local development. It does not contain or encourage hardcoded secrets.
[COMMAND_EXECUTION]: All documented shell commands (e.g., curl for API interaction, npm/pip for installations) are standard development tools used for their intended purpose in managing vendor infrastructure.
[DATA_EXFILTRATION]: No suspicious network activity or sensitive data access was detected. All network communication described is related to the official LaunchDarkly API and documentation.
[PROMPT_INJECTION]: The instructions do not contain any patterns attempting to bypass safety filters or override agent constraints. The focus is entirely on legitimate software architecture migration.

aiconfig-migrate