The Agent Skills Directory

[SAFE]: The skill conducts a static analysis of project files to provide optimization recommendations for AI answer engines. Its operations are transparent and restricted to the project's working directory.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a theoretical attack surface for indirect prompt injection as it processes untrusted user content (e.g., website source code, markdown files) to generate audit findings.
Ingestion points: Project source files are analyzed via glob and read operations in Step 5.
Boundary markers: None explicitly defined within the skill instructions.
Capability inventory: The skill has file-write permissions for report generation and utilizes authorized MCP tool calls.
Sanitization: None specified for input content; reliance on the LLM's inherent processing safety.
[EXTERNAL_DOWNLOADS]: The skill instructions provide a command for the user to install the @upstash/context7-mcp tool. Upstash is a well-known technology provider, and the reference is for optional feature enhancement.
[COMMAND_EXECUTION]: The skill interacts with the platform's Model Context Protocol (MCP) by invoking mcp__claude_ai_Context7__resolve-library-id. This is a legitimate use of the platform's tool-calling capabilities to access updated documentation.

geo-audit