The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local script tools/extract_paper_style.py using the Bash tool to process structural layout references. This script is a vendor-provided resource for style extraction.
[COMMAND_EXECUTION]: Instructions guide the user to run bash tools/install_aris.sh for dependency management if required tools are missing. These scripts are managed by the skill author's infrastructure.
[EXTERNAL_DOWNLOADS]: The skill utilizes WebFetch and WebSearch tools to retrieve paper content from external sources such as arXiv IDs and URLs when users provide a style reference. This data is used solely for structural guidance.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted data from research narratives and external papers to generate its reports.
Ingestion points: Reads NARRATIVE_REPORT.md, STORY.md, AUTO_REVIEW.md, and external paper sources via web retrieval.
Boundary markers: The instructions do not define explicit boundary markers or delimiters for the ingested data.
Capability inventory: Uses Bash execution, Write access to the project root for creating PAPER_PLAN.md and GAP_REPORT.md, and WebFetch for remote data retrieval.
Sanitization: No specific content sanitization or filtering logic is described for the processed text.
[DATA_EXPOSURE]: The skill accesses project-specific files including experiment logs (JSON) and narrative documents to synthesize claims and evidence for the paper outline.

paper-plan