The Agent Skills Directory

[PROMPT_INJECTION]: The skill is designed to ingest and synthesize information from external research repositories (arXiv, Semantic Scholar, OpenAlex). This creates a surface for indirect prompt injection, where instructions embedded in academic papers or metadata could influence the agent's synthesis or subsequent actions.
Ingestion points: Data is retrieved via search scripts search_semantic_scholar.py, search_openalex.py, and search_arxiv.py as defined in SKILL.md.
Boundary markers: While the instructions in SKILL.md and references/dialogue-prompts.md require grounding and citations, there are no explicit boundary markers or instructions to ignore potential commands embedded within retrieved documents.
Capability inventory: The skill possesses the capability to execute local scripts and generate structured documents based on external input.
Sanitization: There is no evidence of sanitization or filtering of the retrieved text before it is processed by the agent.
[COMMAND_EXECUTION]: The skill workflow involves executing local Python scripts (e.g., search_semantic_scholar.py) to perform database queries. These scripts are executed via the shell with arguments generated based on the research topic and dialogue, which is a standard but noteworthy capability.

literature-review