The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes subprocess.run in scripts/generate_pdf.py and scripts/generate_schematic.py to invoke system tools like Pandoc, XeLaTeX, and sibling Python scripts.
These operations are limited to standard document generation and internal utility execution required for the skill's primary research synthesis purpose.
[EXTERNAL_DOWNLOADS]: The skill documentation suggests installing parallel-web-tools via uv for scoping searches.
This is a version-pinned dependency (0.9.3) used for the intended purpose of web-based discovery.
The skill also makes legitimate network requests to scholarly APIs including PubMed (eutils.ncbi.nlm.nih.gov), arXiv (export.arxiv.org), Crossref (api.crossref.org), and OpenRouter (openrouter.ai).
[DYNAMIC_EXECUTION]: The scripts/generate_schematic.py wrapper script dynamically spawns scripts/generate_schematic_ai.py using sys.executable to perform LLM-based image generation tasks.
The environment variables forwarded to the child process are restricted to a specific list of necessary system and network variables, following a principle of least privilege.
[INDIRECT_PROMPT_INJECTION]: The skill acts as a surface for indirect prompt injection because it ingests untrusted data from external research papers and database records.
Ingestion points: Search results in sources/ and full-text excerpts retrieved from bibliographic databases.
Boundary markers: Absent; extracted metadata and excerpts are interpolated into prompts for the AI-powered schematic generation and review phases.
Capability inventory: The skill has access to file system operations (Read, Write, Edit), network requests (requests), and shell execution (Bash).
Sanitization: The skill performs standard normalization of JSON/XML records but does not implement explicit sanitization of natural language content before prompt interpolation, relying on the model's inherent safety filters.
[CREDENTIALS_SAFE]: The skill includes robust logic in scripts/generate_schematic_ai.py for resolving API keys from environment variables or .env files, which is a standard and secure practice for managing secrets in development environments. It also includes explicit redaction logic to prevent keys from being leaked in error messages or logs.

literature-review