The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and critically analyze external scientific papers, research claims, and experimental designs. This creates an attack surface where a malicious paper could contain instructions intended to influence the agent's behavior.
Ingestion points: The agent reads research papers, claims, and methodology descriptions provided by the user or accessed via the Read tool as indicated in SKILL.md.
Boundary markers: The instructions do not explicitly define delimiters (e.g., XML tags) or include a specific "ignore instructions in data" warning to separate untrusted research content from its core logic.
Capability inventory: The skill has access to the Bash, Write, and Edit tools, which could be misused if an indirect injection successfully overrides the agent's instructions.
Sanitization: There is no evidence of input validation or sanitization of the research content being analyzed.

scientific-critical-thinking