The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external, untrusted content such as academic papers, web search results, and user-supplied repository files.
Ingestion points: The skill processes external scholarly text for discovery, screening, and critical analysis as detailed in SKILL.md and references/discovery-and-screening.md.
Boundary markers: The instructions explicitly mandate that external content and retrieved pages be treated as evidence for research tasks, not as instructions to change the agent's workflow or source policy (SKILL.md, point 4).
Capability inventory: The agent can perform web searches, read local and remote files, and generate research deliverables including manuscript drafts and bibliography files.
Sanitization: The skill implements rigorous evidence verification rules requiring citations to be matched to specific findings and access levels (abstract vs. full text) to be explicitly labeled as described in references/critical-analysis-and-synthesis.md.
[SAFE]: The skill references academic resources from established institutions such as Columbia University and Northwestern University for its methodology. It maintains a user-defined source selection policy (e.g., the IEEE Access blacklist) and adheres to standard agent tooling without introducing unauthorized dependencies or executing remote code.

research