The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes data from external scientific sources such as patents, academic papers, and clinical trial results, which represents a potential surface for indirect prompt injection if those sources were to contain adversarial instructions.
Ingestion points: External data enters the context through multiple tools including mcp_virtual-mcp-pharma-intelligence__ls_paper_search, mcp_virtual-mcp-pharma-intelligence__ls_patent_fetch, and mcp_virtual-mcp-pharma-intelligence__ls_clinical_trial_result_search as defined in SKILL.md.
Boundary markers: The prompt instructions do not specify explicit delimiters or 'ignore' instructions to wrap external content.
Capability inventory: The agent's capabilities are primarily restricted to information retrieval and reporting; the skill does not request or utilize tools for file system writing, arbitrary shell command execution, or non-whitelisted outbound network requests.
Sanitization: No specific filtering or sanitization protocols for the retrieved textual data are described in the workflow.
[SAFE]: The skill follows professional standards for pharmaceutical research, employing structured workflows and multi-dimensional scoring systems based on legitimate data sources. All referenced MCP tools are consistent with the vendor's (patsnap) stated domain expertise.

drug-potential-assessment