Remove signs of AI-generated writing from text. Use when editing or reviewing text to make it sound more natural and human-written. Based on Wikipedia's comp...
Security Analysis
high confidenceThis is an instruction-only text-editing skill whose requested capabilities and absence of credentials/install steps are consistent with its stated purpose, but there are minor metadata inconsistencies and one operational caution about file access and invented content.
The name, description, and runtime instructions all describe a text editing/humanization tool and require no external credentials or installs, which is coherent. Minor inconsistencies in metadata exist: SKILL.md lists version 2.1.1 while registry shows 2.1.2 and _meta.json contains ownerId and version values that don't match the registry metadata—this could be an editing/publishing oversight but is worth verifying.
SKILL.md focuses on scanning and rewriting provided text and preserving voice. It does not instruct the agent to read system credentials or unrelated files. However, the allowed-tools list (Read, Write, Edit, Grep, Glob) would permit the agent to search and edit files in the workspace if invoked that way. Also, some guidance ("have opinions", "add soul") encourages inventing voice and may lead the agent to fabricate specific claims, quotes, or attributions if the user doesn’t constrain it. Recommend instructing the agent to limit operations to user-provided text and to avoid inventing factual claims unless explicitly allowed.
No install spec, no code files, and no downloads. As an instruction-only skill this has a low install risk (nothing is written to disk by an installer).
The skill declares no required environment variables, credentials, or config paths. Nothing requests access to external services or secrets, which is proportionate for a text-editing skill.
always is false and the skill is user-invocable; it does not request persistent presence or system-wide changes and does not attempt to modify other skills. Autonomous model invocation is allowed by platform default but not a unique privilege here.
Guidance
This skill is coherent with its purpose and doesn't request credentials or install code, but check a few things before installing: 1) Metadata mismatch: verify the author/owner and version (SKILL.md, registry metadata, and _meta.json disagree) to ensure you're getting the intended package. 2) Limit scope: when invoking the skill, explicitly supply only the text you want edited and instruct the agent not to scan or modify other files in your workspace—allowed tools could otherwise be used to read/search files. 3) Review outputs for invented facts: the skill's "add soul" guidance can lead to fabricated quotes, attributions, or specifics; do not rely on it to produce factual claims without verification. 4) Be aware of the tip link in SKILL.md/README; it points to an external payment page (not code), so evaluate whether you're comfortable with that external link. If you want stronger assurance, ask the publisher to resolve the metadata inconsistencies or run the skill in a sandboxed workspace first.
Latest Release
v2.1.2
Added tip.md badge
More by @oyi77
Published by @oyi77 on ClawHub
