The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions direct the agent to use the Bash tool to run commands for creating visuals and compiling LaTeX. Specifically, it references python scripts/generate_schematic.py and python scripts/generate_image.py, which are dependencies not included in the skill package.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection due to its core function of researching and processing untrusted external data.
Ingestion points: Untrusted research data enters the context during the initial outline and key point identification phase.
Boundary markers: The skill fails to define clear delimiters or provide instructions to ignore embedded commands within the research material.
Capability inventory: The skill operates with high-privilege tools including Read, Write, Edit, and Bash.
Sanitization: No mechanisms for sanitizing or validating external content are described in the writing process instructions.

scientific-writing