The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides examples of curl and ldcli commands for interacting with the LaunchDarkly API. These are used for legitimate configuration tasks as described in the skill's purpose.
[DATA_EXFILTRATION]: The skill uses curl to send configuration data to launchdarkly.com. This is directed at the vendor's own API for the intended functionality of the skill.
[CREDENTIALS_UNSAFE]: The skill references the environment variable $LD_API_KEY for authentication. This is a secure practice for managing secrets in automated environments and does not involve hardcoding credentials.
[PROMPT_INJECTION]: The skill includes an 'Indirect Prompt Injection' surface where user-provided input (e.g., product names) is interpolated into prompts or messages within the created AI configurations. Standard boundary markers like {{product_name}} are suggested in examples to separate data from instructions.

aiconfig-create