The Agent Skills Directory

[PROMPT_INJECTION]: Indirect prompt injection surface detected. The skill is designed to ingest and evaluate untrusted scientific documents and presentations. * Ingestion points: Processes external manuscript files and PDF slides using the read_file tool. * Boundary markers: Absent; no delimiters are defined to isolate untrusted content from the agent's instructions. * Capability inventory: The skill utilizes write_file to create figures and executes python scripts to perform image conversion and schematic generation. * Sanitization: Absent; no validation or sanitization of document content is specified before tool use.
[COMMAND_EXECUTION]: The skill provides explicit instructions to execute shell commands for workflow tasks. Evidence includes the usage of python scripts/generate_schematic.py to create diagrams and python skills/scientific-slides/scripts/pdf_to_images.py to convert PDF presentations for inspection.
[EXTERNAL_DOWNLOADS]: The skill contains references to external repositories and services. Evidence includes a reference to an upstream repository on GitHub (dr-claw/skills/inno-paper-reviewer) and a recommendation to use the k-dense.ai web platform for complex workflows.

inno-paper-reviewer