Span-level replay: fork from the failure
Fork-replay one recorded span from a failed agent run and get an attribution verdict.
我該用這個嗎
品質與安全性
A
根據工具定義與協定合規性的自動化分析。
上下文成本
~119Token（工具定義）
~780 B典型回應大小
極小的注意力影響（128k 上下文的 0.09%）
這是每次將伺服器的工具載入模型上下文時所消耗的約略 token 數量。數量越高，可用於其他工作的注意力就越少。
安裝
一鍵安裝
將以下內容加入你的 `claude_desktop_config.json` 檔案：
{
"mcpServers": {
"span-level-replay-fork-from-failure": {
"url": "https://textforagents.com/products/span-level-replay-fork-from-failure/mcp"
}
}
}遠端端點
https://textforagents.com/products/span-level-replay-fork-from-failure/mcpstreamable-http它能做什麼
工具清單
工具（1）
🟢 唯讀🟡 寫入🔴 刪除⚪ 未知
⚪replay_span(span, span_id, spans)
Fork-replay ONE recorded span from a failed agent run and return an attribution verdict (determinism vs tool drift vs env drift vs world-state vs model decision), without re-running the whole run.
輸入結構描述
{
"type": "object",
"properties": {
"span": {
"type": "object",
"description": "One normalized span record."
},
"span_id": {
"type": "string"
},
"spans": {
"type": "array",
"items": {
"type": "object"
},
"description": "A recorded run's span records; span_id selects the one to fork."
}
}
}社群
證據
近期觀測
已驗證未記錄版本1 個工具