The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes shell and Python snippets in SKILL.md and references/review-pipeline.md for automated text processing tasks, such as dash normalization and cross-referencing markers across different markdown files.- [EXTERNAL_DOWNLOADS]: In references/community-research.md, the skill provides instructions and code snippets for using browser automation and search tools to fetch data from well-known platforms including Reddit, Hacker News, and GitHub for research purposes.- [PROMPT_INJECTION]: The skill identifies a surface for indirect prompt injection by ingesting data from external community forums (Category 8). It explicitly defines an evidence chain to mitigate this risk: 1. Ingestion points: Reddit, Hacker News, and GitHub Issues content (defined in community-research.md). 2. Boundary markers: A mandatory fact-checking table used to distinguish between user-provided data and external findings (defined in per-section-drafting.md). 3. Capability inventory: Local script execution for file normalization and style auditing. 4. Sanitization: Detailed instructions for verifying external data and providing source links for all findings.

living-text-style