The Agent Skills Directory

[SAFE]: The skill uses a local Python script scripts/run.py to analyze the project workspace. The script identifies missing files, duplicate sentences, and stylistic 'smells' (e.g., repetitive openers) using regex patterns.
[SAFE]: File access is restricted to the workspace directory, specifically targeting markdown files in sections/ and configuration JSONL files in outline/ and citations/.
[SAFE]: The script performs only read and write operations on local files within the workspace. It generates a summary report in output/WRITER_SELFLOOP_TODO.md and logs quality issues through a local project module.
[SAFE]: No network operations, external downloads, or execution of untrusted code were identified. The imports used in the script (tooling.common, tooling.quality_gate) are internal project dependencies.
[SAFE]: A hex escape (\x27) is used in SKILL.md within the phrase subsection\x27s, which decodes to a standard single quote (') and is entirely benign.

writer-selfloop