The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses several shell commands for document validation and artifact generation. Specifically, it executes pdfinfo, pdffonts, grep, and file to verify LaTeX output properties (page count, broken references, embedded fonts, and figure quality). Additionally, it uses pdflatex to compile TikZ-based diagrams during the figure synthesis workflow.
[INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it is designed to ingest and process untrusted external data, including meeting notes, research drafts, source code, and related work PDFs.
Ingestion points: Data enters the context via the story_mode.md workflow (notes, drafts, code, reviews) and the figure_synthesis_guide.md spec process.
Boundary markers: The skill does not explicitly define structural delimiters or 'ignore' instructions for the processed content.
Capability inventory: Capabilities include local shell execution for validation/compilation and network access through the Gemini API for image generation.
Sanitization: No explicit sanitization or escaping of external content is mentioned before it is interpolated into agent prompts.
[EXTERNAL_DOWNLOADS]: The skill documentation describes installing the google-genai Python package via pip to enable AI-powered figure generation. It also references cloning the skill repository from GitHub.

paper-writing