The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests feature flag configurations from the LaunchDarkly API via the get-flag tool. These configurations include user-controllable fields like flag descriptions and rule names that could be used to host malicious instructions intended to influence the agent's behavior during a targeting change workflow.
Ingestion points: Flag state and metadata are retrieved in SKILL.md using the get-flag MCP tool.
Boundary markers: The skill does not define specific delimiters or instruction-ignore warnings to wrap flag metadata when it is presented to the agent's context.
Capability inventory: The skill possesses significant modification capabilities including toggle-flag, update-rollout, and update-targeting-rules as described in SKILL.md and references/targeting-patterns.md.
Sanitization: No sanitization or filtering of the API responses is performed before the data is used to inform targeting decisions.
[EXTERNAL_DOWNLOADS]: The skill references and depends on the official LaunchDarkly MCP server and documentation repositories.
Evidence: marketplace.json requires the @launchdarkly/mcp-server and README.md provides links to the vendor's official GitHub repository and documentation site.

launchdarkly-flag-targeting