The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses git (checkout, commit, push) and gh (pr create, api) CLI tools to automate the workflow of enabling new skills. These operations are scoped to creating branches and opening PRs, providing a human-in-the-loop safety checkpoint.- [EXTERNAL_DOWNLOADS]: The Analyze mode fetches content from arbitrary user-provided URLs using WebFetch to extract metadata, RSS feeds, and social handles. This is the primary mechanism for generating recommendations.- [PROMPT_INJECTION]: By processing content from external websites, the skill is subject to Indirect Prompt Injection. The instructions explicitly direct the agent to check for and mitigate 'ignore previous' instructions or role-play attempts within the fetched content, reducing the risk of malicious site content hijacking the recommendation logic.

auto-workflow