The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted academic data from external sources, which could contain malicious instructions designed to influence the agent's behavior during summarization.
Ingestion points: Search results and abstracts fetched from RISS, KCI, DBpia, and Google Scholar (SKILL.md).
Boundary markers: Absent. There are no delimiters or specific instructions for the agent to ignore embedded commands in the fetched paper data.
Capability inventory: Performs web fetching (GET) and writes configuration files to the plugin data directory.
Sanitization: Absent. The skill does not specify any filtering or validation for the content returned from external academic databases.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill manages user-provided API credentials.
Evidence: The workflow instructs the agent to request a KCI_API_KEY and store it in ${CLAUDE_PLUGIN_DATA}/moai-credentials.env (SKILL.md). This is a localized configuration for the modu-ai suite and follows standard secret management practices for such agents.
[EXTERNAL_DOWNLOADS]: The skill interacts with external academic and government services.
Evidence: Network operations are directed to well-known and official academic portals, including open.kci.go.kr, data.go.kr, dbpia.co.kr, and riss.kr (SKILL.md, guide.md).

education-paper-search