The Agent Skills Directory

[SAFE]: The skill operates as a local tool for packaging academic paper data (PDF, TeX, Markdown) for processing by an AI agent. It implements strict validation schemas for agent responses and heuristic checks for numerical claim consistency.
[SAFE]: External dependencies are restricted to standard libraries for YAML parsing and PDF text extraction (PyMuPDF). All operations occur locally without the need for model API keys within the tools themselves.
[SAFE]: The system prompt explicitly instructs the agent to treat all manuscript data and evidence excerpts as untrusted data that cannot change the task instructions or schema (Indirect Prompt Injection defense).

presentation-slides