The Agent Skills Directory

[COMMAND_EXECUTION]: The skill defines a revision mode that utilizes local Python scripts (ars_anchorize_draft.py and ars_apply_revision_patch.py) to manage manuscript updates through a block-based hashing and patching system. This mechanism is designed to prevent silent distortion of text during AI-assisted revisions.
[COMMAND_EXECUTION]: The visualization_agent is instructed to generate publication-quality Python (using matplotlib and seaborn) and R (using ggplot2) code for data visualization. This code is intended to be reviewed and executed by the user or an orchestrator to produce figures for the academic paper.
[EXTERNAL_DOWNLOADS]: The literature_strategist_agent and citation_compliance_agent reference established academic databases and tools (e.g., ERIC, Scopus, PubMed, and Retraction Watch) for source verification and literature search. All referenced domains are well-known academic services.
[SAFE]: The skill includes an 'Anti-Leakage Protocol' (Knowledge Isolation) which instructs writing agents to prioritize provided session materials over their parametric memory. This prevents the fabrication of methodology and ensures factual accuracy based on user-provided research.
[SAFE]: The instructions for the agents are highly structured and focused on academic standards (APA 7th, CRediT authorship, etc.), with explicit checkpoints for user approval and transparency statements regarding AI usage.

academic-paper