The Agent Skills Directory

[SAFE]: The skill stores user-provided business context in local files to maintain persistence across sessions. It targets specific directories associated with development tools, such as ~/.claude/ and .cursor/rules/. The agent is explicitly instructed to confirm the exact file path with the user before performing any write operations, ensuring user oversight and preventing unauthorized file access.
[SAFE]: The skill ingests external data from Google search results to perform research. Although this is a common vector for indirect prompt injection, the skill defines highly restrictive output requirements (such as specific banned vocabulary lists and structural templates) that prioritize user-defined quality standards over external instructions found in the researched content.
[SAFE]: The skill provides numerous references and citations to legitimate SEO research and open-source writing tools. All external links point to established academic, industry, or reputable community resources.

write-content