The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the installation of the sciverse-mcp-server package from the NPM registry. This package acts as the bridge between the agent and the Sciverse API, facilitating all paper retrieval tasks.
[COMMAND_EXECUTION]: The setup instructions include commands for npm install and the use of npx within a configuration file to execute the sciverse-mcp-server. These commands are standard for initializing and running the required MCP server environment.
[INDIRECT_PROMPT_INJECTION]: The skill operates by retrieving and processing text from external academic papers through tools like semantic_search and read_content. This ingestion of third-party data represents a standard attack surface for indirect prompt injection, where malicious instructions could be embedded in the scientific literature.
Ingestion points: External text is brought into the agent's context via the read_content and semantic_search tools described in SKILL.md.
Boundary markers: The instructions do not specify explicit delimiters or "ignore" instructions for the retrieved content.
Capability inventory: The skill includes tools for searching metadata, semantic retrieval, reading content byte-ranges, and fetching images.
Sanitization: There is no mention of sanitization or filtering of the retrieved academic text before it is presented to the agent.

sciverse