The Agent Skills Directory

[SAFE]: The skill stores all user-specific configurations, such as portfolio properties and Typefully workspace IDs, in the local file system (e.g., ~/.config/makerskills/jab-hook/). This ensures that sensitive personal data is managed locally and is not hardcoded or leaked through the skill's instructions.- [SAFE]: Interaction with external services like Typefully is performed via Model Context Protocol (MCP) tools (mcp__typefully__*). This is a secure, authenticated method for AI agents to interact with third-party APIs.- [SAFE]: The "inspiration scan" feature, which involves scraping content from LinkedIn and X using tools like agent-browser, is designed for structural analysis. The instructions explicitly direct the agent to extract formatting patterns and structural archetypes rather than interpreting or obeying the content of the scraped posts, which mitigates risks from untrusted data ingestion.- [SAFE]: The skill does not include any executable scripts, binary files, or remote code downloads. It relies entirely on structured instructions and native agent capabilities, adhering to the principle of least privilege.

jab-hook