The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes several Bash scripts (scripts/*.sh) that automate article management tasks such as creating directories (mkdir), searching for files (find), and counting words using standard text utilities (sed, grep, tr, wc). These operations are local to the project environment.
[EXTERNAL_DOWNLOADS]: The documentation references the installation and use of @hocgin/wenyan-cli, an NPM package used for AI-driven image generation. As this package belongs to the skill's author ('hocgin'), it is considered a legitimate vendor resource for the skill's extended functionality.
[COMMAND_EXECUTION]: The scripts/text-audit.sh script executes an embedded Python script to perform offline analysis of article drafts (e.g., checking sentence length and word density). This script runs locally and does not perform network operations or access sensitive system configuration files.
[PROMPT_INJECTION]: The skill's research module (rules/research.md) involves processing untrusted data from external URLs and PDFs, which presents a surface for indirect prompt injection.
Ingestion points: Data enters the context through web fetching and PDF reading tools.
Boundary markers: The skill uses Markdown structures (headers and code blocks) to organize findings, though it lacks explicit 'ignore instruction' delimiters for processed content.
Capability inventory: The skill has access to file writing, shell execution, and network tools.
Sanitization: The skill provides instructions for the agent to extract and structure data into reports, which acts as a manual sanitization process through the model's summarization.

article-writer