text-humanizer-Instruction-based

Paste the text and ask to humanize, de-AI, or rewrite it. The Skill scans for 45 documented AI writing patterns—buzzwords, filler phrases, and structural tells—and applies a specific rewrite rule for each one found, returning clean prose.

It covers 45 patterns in three categories: vocabulary tells like "delve" and "leverage," phrase-level tells like "It's worth noting that," and structural tells like em-dash overuse, forced three-item lists, and sycophantic openers such as "Great question!"

No. The Skill performs all detection and rewriting using the agent's own language capability against a fixed pattern list. There are no external calls, no API keys, and no dependencies to install.

Yes. Ask for a detection report to get a flagged list of pattern codes with the offending phrases quoted, followed by the rewrite. Audit mode instead annotates the original text inline so you can learn which patterns to avoid.

No. The Skill only changes style, never factual content. Its guardrails explicitly prohibit altering claims, adding hedges, or fabricating examples—if a sentence needs a specific example, it inserts a placeholder and notes it.

Avoid it for code, technical documentation, translations, and text you wrote yourself that only needs grammar fixes. It is also not designed to evade AI detection tools—it targets writing quality, not detection evasion.