The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest untrusted data from external and local sources, which creates a potential surface for indirect prompt injection.
Ingestion points: The workflow in SKILL.md instructs the agent to "Search the web for recent methodological papers, systematic reviews, and research design innovations" and "Read existing project files (topic proposals, literature reviews, course requirements, advisor feedback) for context."
Boundary markers: The instructions do not define any specific delimiters or instructions to isolate the data being read, making the agent potentially susceptible to following instructions embedded within external research materials or project documents.
Capability inventory: The skill utilizes web search, file reading, and file writing capabilities (writing deliverables such as {project}-research-design.md to the project directory) as described in its agentic protocol.
Sanitization: There is no evidence of content validation or sanitization protocols to verify the integrity of the information retrieved from web searches or local files before it is processed by the agent.

alterlab-rma-research-designer