The Agent Skills Directory

[COMMAND_EXECUTION]: The skill orchestrates a series of maintenance, validation, and testing tasks by executing local scripts using bun, node, and bash. These include internal scripts for cache management (e.g., .claude/scripts/update-docs-cache.ts) and plugin validation tools (claude plugin validate).
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process potentially untrusted external data such as GitHub tickets, pull request bodies, and external web pages. It proactively mitigates the risk of indirect prompt injection by utilizing <pasted_content> delimiters and explicit instructions to the AI model to treat such content as task data rather than directive instructions.
[DYNAMIC_EXECUTION]: To achieve parallel analysis of different marketplace components, the skill dynamically spawns and coordinates multiple specialized sub-agents (e.g., optimizer-skills, optimizer-commands) using the Agent tool.
[EXTERNAL_DOWNLOADS]: The skill performs network operations to maintain an up-to-date documentation cache from remote sources. It also references well-known external services for MCP (Model Context Protocol) configurations, such as official Nuxt documentation endpoints.

marketplace-optimizer