The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill employs a two-stage writing process that ingests external data from the research-lookup tool to generate manuscript outlines and prose.
Ingestion points: The agent is instructed to use research-lookup to gather literature and data from external sources.
Boundary markers: There are no specific delimiters or instructions provided to the agent to treat external research content as untrusted or to ignore embedded instructions within that content.
Capability inventory: The skill uses the Bash tool to execute Python scripts and LaTeX compilation, and the Write tool to create manuscript files.
Sanitization: The instructions do not specify any validation or sanitization steps for the data fetched via the research tool before it is processed into the final manuscript.
[COMMAND_EXECUTION]: The skill instructions direct the agent to utilize the Bash tool for several tasks related to its primary function.
Evidence: The agent is instructed to execute local Python scripts (python scripts/generate_schematic.py and python scripts/generate_image.py) to create figures and graphical abstracts.
Evidence: The skill requires the compilation of LaTeX documents using system tools like xelatex and bibtex to produce professional reports.
[DYNAMIC_EXECUTION]: The skill facilitates the generation of LaTeX source files (.tex) based on templates and user input, which are subsequently compiled into executable or viewable formats (PDF).
Evidence: The scientific_report_template.tex and scientific_report.sty files are provided as a basis for dynamically generating and compiling complex scientific documents.

scientific-writing