The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes two local TypeScript scripts (scripts/sm2.ts and scripts/scan-signals.ts) to handle spaced-repetition scheduling and learner profile adaptation. These scripts are designed for Node.js (v22+) or Bun and perform standard filesystem operations within a dedicated local directory (~/.explain-this/). They do not utilize external dependencies or execute arbitrary shell commands.
[DATA_EXFILTRATION]: The skill manages a 'Learner Profile' and study history stored in ~/.explain-this/. Analysis of the provided scripts confirms that data is processed strictly locally; there are no network-capable functions (e.g., fetch, curl, or socket operations) present in the logic. The README explicitly advises the user that this directory contains personal data and should not be committed to repositories.
[EXTERNAL_DOWNLOADS]: The skill does not perform any remote downloads or package manager operations (npm/pip). It relies on the host environment having Node.js or Bun pre-installed.
[PROMPT_INJECTION]: The skill uses detailed instructional guidelines to enforce its educational methodology, such as requiring a learner interview before explaining content. These constraints are functional rather than malicious and do not attempt to bypass agent safety filters or extract system prompts.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and explain external artifacts (PDFs, code, articles). While this presents a standard ingestion surface for agent skills, the provided adapters and quiz generation logic focus on structured analysis and do not include capabilities that would allow for high-risk exploits from malicious content within those artifacts.

explain-this