The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to utilize local command-line tools such as grep to perform pedagogical analysis of the codebase. These operations are restricted to the resolved project directory and are used solely to locate examples for tutorial generation.
[EXTERNAL_DOWNLOADS]: The 'External source' entry point (Entry [f]) allows the agent to fetch content from URLs, such as Apple developer documentation or technical blog posts. This functionality uses standard agent tools for research and learning synthesis.
[PROMPT_INJECTION]: Ingestion points: Entry [f] in SKILL.md (via external URL fetching). Boundary markers: The skill includes a non-negotiable requirement for an 'honesty-machine' section to identify what a source leaves out, forcing a critical analysis of the data. Capability inventory: Local file system writes (tutorials, config, logs) and codebase search via grep. Sanitization: None mentioned. This indirect prompt injection surface is a core component of the skill's learning-to-write methodology and is managed via analytical constraints.
[SAFE]: The skill implements a robust project resolution discovery chain that includes explicit safeguards to prevent the agent from traversing into other users' home directories or beyond the filesystem root. Project configuration and history are maintained in local .claude/ directories.

tutorial-creator