The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its intended workflow of processing external data retrieved via the research-lookup skill into a manuscript draft.
Ingestion points: Untrusted research data enters the agent context from external literature searches facilitated by the research-lookup tool (described in Section 7).
Boundary markers: The instructions do not define clear delimiters or use warnings to ignore instructions potentially embedded within the retrieved literature (SKILL.md).
Capability inventory: The skill instructs the agent to execute Python scripts (scripts/generate_schematic.py, scripts/generate_image.py) and LaTeX compilation commands (xelatex) in the local environment (SKILL.md).
Sanitization: There is no evidence of data sanitization or validation performed on the research findings before they are processed by the agent into the final narrative sections.
[COMMAND_EXECUTION]: The skill directs the agent to execute specific shell commands and scripts to fulfill its primary purpose.
The agent is prompted to execute local Python scripts (scripts/generate_schematic.py and scripts/generate_image.py) to generate graphical abstracts and other figures (Section 1).
The skill provides instructions for the agent to use xelatex to compile LaTeX documents into professional reports (Section 8).

scientific-writing