AI Agent Hub
Back to skills
TencentCloud Text AIGC Detection icon

TencentCloud Text AIGC Detection

Data Analysis Updated 2026.08.29

Paste the following prompt into your AI chat to install this skill:

Follow https://skillhub.cn/install/skillhub.md and install @tencent-adm/tencentcloud-aigc-recog-text.

About this skill

Problem

In moderation, Q&A, and news authenticity workflows, it is hard to tell whether text was written by a human or generated by a large model. Simple keyword rules are fragile, and asking a model to self-assess is not a reproducible control. For engineers, the key point is turning the judgment into auditable return fields rather than informal explanations.

How It Works

The skill wraps TencentCloud TextModeration with TEXT_AIGC recognition. Input can be direct text, a --file path, or piped data via --stdin. It requires TENCENTCLOUD_SECRET_ID, TENCENTCLOUD_SECRET_KEY, and TENCENTCLOUD_AIGC_RECOG_TEXT_BIZ_TYPE, then returns structured fields:
- suggestion: Pass, Review, or Block
- score: confidence from 0 to 100
- label: such as Normal or GeneratedContentRisk
- keywords and detail_results: matched keywords and dimensional details

Boundaries

Each request is limited to 10000 Unicode characters, with 50 requests/second. It is intended for news, encyclopedic, Q&A, comment, and novel-style text, but outcomes depend on the TencentCloud service and the BizType policy. If credentials are missing or the call fails, the skill should surface the error rather than let a model guess the detection result.

Use Cases

  • Editors screen news drafts before publication and flag suspected AI-generated paragraphs for review.
  • Moderation teams read comment files and use Block or Review suggestions to shortlist high-risk posts.
  • QA staff pipe Q&A answers through stdin and inspect score and keywords to decide on manual review.
  • Content operators check encyclopedia drafts and separate human-edited text from bulk-generated text.

Best For

  • Content moderation engineers: need to screen news, comments, and Q&A for suspected AI generation before release.
  • Community risk staff: want to rank suspicious posts by score and prioritize high-risk items for review.
  • Data operators: want to filter bulk-generated entries from text files for cleanup.
  • QA developers: need to call the text AIGC API and parse suggestion, label, and keywords.