The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Downloads the hypogenic package and associated research datasets from the Chicago Human+AI Lab (ChicagoHAI) GitHub organization.
[COMMAND_EXECUTION]: Executes shell commands to install dependencies, clone repositories, and configure the GROBID service for literature processing.
[PROMPT_INJECTION]: The skill contains a vulnerability surface for indirect prompt injection as it processes external datasets and research papers without explicit input sanitization or boundary delimiters in its LLM prompt templates.
Ingestion points: Data is ingested from local JSON files and PDF documents located in directories specified in the configuration.
Boundary markers: Data samples and literature insights are interpolated into prompts using placeholders like {data_samples} and {hypothesis} without surrounding delimiters or instructions to ignore embedded commands.
Capability inventory: The skill can execute shell commands for setup and make API calls to LLM providers.
Sanitization: No content-based filtering or sanitization is applied to the processed research data.

hypogenic