The Agent Skills Directory

[SAFE]: The skill implements a robust protocol for document review, focusing on linguistic clarity metrics such as sentence length, passive voice, and nominalization density.
[SAFE]: Automated checks in scripts/checks.py use deterministic regular expressions to identify grammatical patterns, avoiding unsafe execution of user-supplied text.
[SAFE]: The skill uses a shared library architecture for reporting findings, ensuring consistent and validated JSON output via scripts/merge.py.
[SAFE]: External dependencies are limited to standard libraries like jsonschema, used for output validation.
[SAFE]: The protocol includes a 'Clean-context critique' step, which explicitly instructs the agent to disregard external framing or steering that might accompany an artifact, serving as a defense against prompt injection.

critique-clarity