The Agent Skills Directory

[SAFE]: The skill implements a deterministic documentation pipeline using controlled subprocess calls. The scripts/convert.sh script uses Bash arrays for command construction, which is a secure practice that prevents shell injection when passing arguments to pandoc.
[SAFE]: The scripts/readability.py script computes readability metrics using standard formulas (Flesch-Kincaid, SMOG, etc.) by performing static text analysis. It does not execute or dynamically evaluate the content of the files it processes.
[SAFE]: Indirect Prompt Injection Surface: The skill's workflow requires the agent to read and summarize existing project documentation and codebase comments. While this is an entry point for untrusted data (Category 8), it is essential to the skill's primary purpose. The skill lacks dangerous sinks; ingestion is limited to text analysis and file generation via standard tools.
Ingestion points: SKILL.md (Step 1: Analyze the Project; R2: Detect Existing README).
Boundary markers: Absent.
Capability inventory: Subprocess execution of pandoc via scripts/convert.sh and filesystem write operations for documentation output.
Sanitization: Analysis scripts treat input as non-executable prose.

tech-writer