Humanize AI-generated text to bypass detection. This humanizer rewrites ChatGPT, Claude, and GPT content to sound natural and pass AI detectors like GPTZero,...
Security Analysis
medium confidenceThe skill is internally coherent (it contains detectors and transformers that match its stated goal) but is explicitly designed to help users evade AI-detection tools and has no provenance — ethically risky and worth caution before installing.
The name/description match the code and SKILL.md: scripts detect, transform, and compare AI-style patterns and a patterns.json contains the rules. Nothing requested (no env vars, no installs) is out of scope for the declared purpose. However, the stated objective is explicitly to 'bypass detectors', which is an adversarial use-case (not a technical mismatch, but a red flag for misuse).
SKILL.md instructs batch scanning and automated transformation to 'bypass' detectors and directs editing patterns.json to tune behaviour. The runtime instructions and code operate only on local files and make no external network calls, but the explicit goal (evading detection) is deceptive/abusive. Also note the skill gives broad discretion (aggressive mode, customizable patterns) that can be used to tune evasion at scale.
Instruction-only skill with bundled Python scripts and JSON; no external install or downloads. Nothing writes to unusual system locations. No high-risk install mechanisms detected.
No environment variables, credentials, or config paths are required. The scripts only read/write user-specified files; requested environment access is minimal and proportionate to the stated file-processing purpose.
Skill is not always-enabled and does not request persistent privileges or modify other skills or system-wide settings. It runs as local CLI tooling when invoked.
Guidance
This package appears to do exactly what it says: detect AI-style patterns and rewrite text to reduce those signals. It does not request credentials or phone home, which reduces technical risk. However: (1) its explicit purpose is to help users evade AI/plagiarism detectors — that is an ethically and potentially legally problematic use (academic dishonesty, contract/terms violations, etc.); (2) the source and homepage are unknown, so you cannot verify provenance, maintenance, or licensing; (3) it includes code you should review before using (patterns.json drives detection and can be edited to more aggressively evade detectors); and (4) test it offline on non-sensitive text first because automated rewrites may alter meaning or introduce factual errors. If you intend to use it for legitimate editing/quality-improvement, consider removing or repurposing the 'bypass' language and auditing patterns.json; if you would use it to evade detection in contexts that prohibit that behavior (schools, publishers, legal processes), do not install or use it. Additional useful checks: inspect patterns.json for any unexpected strings, run the scripts in an isolated environment, and confirm there are no network calls if you modify the code.
Latest Release
v1.0.1
humanize-ai-text 1.0.1 - No code changes detected in this release. - Documentation and usage instructions remain unchanged. - Maintains compatibility and functionality as described in the previous version.
Popular Skills
Published by @moltbro on ClawHub
