Ai Writing Humanizer
PassAudited by ClawScan on May 1, 2026.
Overview
This appears to be a simple local prose-rewriting skill with no credential, network, or persistence behavior, but it can automatically change long outgoing text so the result should be reviewed.
This skill looks safe to install from an agentic-security perspective. Before relying on it, review the final prose for meaning changes and make sure using a tool to remove AI-writing cues is appropriate for your context.
Findings (2)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Recipients may read the output as more naturally human-written than the original AI-assisted draft, which can matter in settings with AI-use disclosure rules.
The skill is explicitly designed to remove signs that text may have been AI-generated. This is disclosed and central to the skill, but users should consider disclosure expectations for user-facing communications.
"ensure a natural, human tone without obvious AI tells"
Use it for style cleanup only where allowed, and do not use it to bypass required AI-authorship disclosure.
If pointed at the wrong output path, it could overwrite a file; if used before sending without review, it could change wording or meaning.
The local helper reads and writes files specified by command-line arguments. This matches the documented workflow, but the output file can be overwritten and the resulting text may be changed automatically.
with open(args.input, 'r') as f:
src = f.read()
...
with open(args.output, 'w') as f:
f.write(cleaned)Write to a separate output file and review the cleaned text before sending or publishing it.