Bỏ qua tới nội dung
TakoAPI
Quay lại kỹ năng

social-trust-manipulation-detector

bởi andyxinweiminicloud

Coding Agents & IDEs
251Lượt tải do clawskills.sh công bố
0Lượt xem
0 lượt thíchĐăng nhập để thích kỹ năng này
—Chưa có đánh giá
Đánh giá của bạn

Tóm tắt

Helps identify coordinated social trust manipulation in agent marketplaces — catching reputation gaming.

Mô tả

Kỹ năng này làm gì

Analyzes social trust signals in agent marketplaces to identify when a skill's reputation is built on coordinated manipulation rather than genuine community validation. Detects sockpuppet networks, burst voting patterns, and manufactured engagement that inflate a skill's apparent trustworthiness.

Marketplace reputation systems treat all upvotes equally; this tool analyzes behavioral fingerprints and statistical anomalies that distinguish manufactured consensus from earned trust.

Khi nào nên dùng

  • Auditing a highly-upvoted skill before installing it on a production agent
  • Investigating a publisher whose skills all launched with sudden mass popularity
  • Checking whether a marketplace's top-rated skills reflect real independent opinions
  • Flagging suspicious engagement before recommending a skill to a team
  • Identifying mutual support networks spread across multiple publisher accounts

Quy trình ví dụ

ĐẦU VÀO

User asks: Assess social trust integrity for the `ai-assistant-toolkit` publisher

AGENT
  1. 1

    Collect engagement velocity data for all skills published under the account

  2. 2

    Analyze account creation dates and cross-voting patterns among early upvoters

  3. 3

    Compare each skill's upvote-to-install ratio against organic baselines for similar skills

  4. 4

    Detect cross-publisher coordination by checking upvoter overlap with other publisher networks

  5. 5

    Assess top reviews for linguistic diversity and specificity versus template-generated phrasing

ĐẦU RA

Manipulation verdict MANUFACTURED: all four skills show coordinated burst voting, sockpuppet cohort fingerprints, upvote-to-install ratios 4-18x above organic baseline, and a mutual support network with two other publishers

Cài đặt

clawhub install andyxinweiminicloud/social-trust-manipulation-detector

Chạy lệnh CLI, hoặc dán prompt vào Claude Code, Cursor, Codex hay coding agent bạn đang dùng.

Liên kết