The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute local Python scripts (scripts/generate_schematic.py and scripts/generate_image.py) using the Bash tool to create visuals for the manuscript. Because these scripts are referenced but not included in the provided file set, their logic cannot be audited for security.
[PROMPT_INJECTION]: The skill instructions emphasize mandatory figure generation and prose formatting using 'MANDATORY' and 'CRITICAL' tags. While these serve to enforce a specific persona and quality standard, they use forceful language common in system prompt overrides.
[PROMPT_INJECTION]: There is an inherent risk of indirect prompt injection as the skill is designed to fetch and process external literature via the research-lookup tool. Untrusted data from scientific papers could contain instructions intended to influence the agent's output.
Ingestion points: External data enters the context through research queries as described in the two-stage writing process in SKILL.md.
Boundary markers: The instructions do not define specific delimiters or guidelines to ignore instructions embedded in the retrieved research content.
Capability inventory: The skill has access to the Bash tool (for script execution) and Write/Edit tools (for file modification).
Sanitization: There is no explicit requirement for the agent to sanitize or validate the content fetched from external databases.

scientific-writing