ClawHub

Verified Humanizer

v0.0.7

Transform AI-generated content into natural, human-sounding writing, measure the improvement, and optionally verify the result.

⭐ 0· 97·0 current·0 all-time
by@nutstrut
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Benign
high confidence
✓
Purpose & Capability
Name/description (humanizing AI text, measuring improvements, optional verification) aligns with provided files: rewrite guidance (SKILL.md, README), evaluation templates, local helper scripts, and lightweight OpenClaw hook. There are no unrelated dependencies, credentials, or binaries requested.
✓
Instruction Scope
SKILL.md explicitly requires local rewriting, local evaluation, and that verification (if used) only transmit structured metrics — and the included scripts/hooks follow that guidance. No instructions request reading unrelated system files, secrets, or sending raw text externally.
✓
Install Mechanism
No install spec is provided (instruction-only style). The package includes small local scripts and hook files; nothing downloads or executes external code or archives. This is low-risk and proportionate to the stated purpose.
✓
Credentials
No environment variables, credentials, or config paths are required. The skill does not ask for unrelated secrets and the SKILL.md forbids sending private text to verifiers, so requested access is minimal and appropriate.
✓
Persistence & Privilege
always is false and the hook only returns a brief system reminder (and skips when context.subagent is present). The skill does not request persistent/global privileges or modify other skills' config.
Assessment
This package appears coherent and privacy-focused: it keeps rewriting and evaluation local and does not request credentials or external services. Before installing, verify that any external verification you choose to integrate (if you enable verification later) is configured to accept only structured metrics (no raw or rewritten text). Also review and sandbox the included hook scripts so the system message they inject matches your agent policy before you enable the skill in production.

Like a lobster shell, security has layers — review code before you run it.

latestvk970cc6p3d4hdcnxt61s7x0k3983twg6

License

MIT-0
Free to use, modify, and redistribute. No attribution required.
Termshttps://spdx.org/licenses/MIT-0.html

Comments