{
 "title": "io.github.GuardBee/toxic-flow-auditor",
 "site": "台灣 AI 代理人任務中樞",
 "canonical": "https://penguindriver.com/s/io.github.GuardBee%2Ftoxic-flow-auditor",
 "markdown": "https://penguindriver.com/s/io.github.GuardBee%2Ftoxic-flow-auditor.md",
 "summary": "Finds lethal-trifecta toxic flows in MCP tool catalogs: untrusted input, sensitive data, egress",
 "date_modified": null,
 "retrieved": "2026-10-10",
 "content_sha256": "1cce25775a6d668dd740311769b291911e2a2d8c9f4eae3a568ba47709da23de",
 "fingerprint_basis": "Markdown 版全文，移除贊助行與開頭 front matter，去除頭尾空白",
 "verify": {
  "text_snapshot": "https://penguindriver.com/cite?path=%2Fs%2Fio.github.GuardBee%252Ftoxic-flow-auditor&format=text",
  "how": "下載 text_snapshot 後計算 SHA-256，應與 content_sha256 相同；內容更新時指紋會改變"
 },
 "sources": [
  {
   "name": "https://github.com/GuardBee/guardbee-mcp",
   "url": "https://github.com/GuardBee/guardbee-mcp"
  }
 ],
 "source_count": 1,
 "cite_as": {
  "zh": "io.github.GuardBee/toxic-flow-auditor｜台灣 AI 代理人任務中樞。https://penguindriver.com/s/io.github.GuardBee%2Ftoxic-flow-auditor（資料日期 未標示，擷取 2026-10-10）",
  "apa": "台灣 AI 代理人任務中樞. (2026). io.github.GuardBee/toxic-flow-auditor. Retrieved 2026-10-10, from https://penguindriver.com/s/io.github.GuardBee%2Ftoxic-flow-auditor",
  "markdown": "[io.github.GuardBee/toxic-flow-auditor](https://penguindriver.com/s/io.github.GuardBee%2Ftoxic-flow-auditor)（台灣 AI 代理人任務中樞，2026-10-10）"
 },
 "license": "可引用，請附上正式網址與資料日期"
}