context-budget-analyzer@1.5.0
VERIFIEDPublisher claims: "Agent 上下文预算分析器。扫描一次 Agent 会话的对话记录、Transcript 或手工提供的 上下文清单,按"系统指令 / 任务目标 / 约束 / 工具输出 / 对话历史 / 中间推理 / 状态与记忆"分类统计 Token 占用,识别上下文污染与 Compaction 风险: 超长工具输出、重复粘贴内容、失效假设、约束丢失风险,并给出分层治理建议 (原文保留 / 结构化 / 留指针 / 可丢弃)。约束审计会解析压缩摘要行,对照 判定每条约束是"被压缩丢弃"还是"被摘要救回",直答"压缩丢掉了什么"。 触发词:分析上下文,上下文预算,Token 都花在哪了,上下文为什么这么长, 压缩会丢什么,context budget,analyze context,context usage, token breakdown,compaction risk,会话太长了,上下文污染, 上下文增长,usage 轨迹,context growth,compaction 检测, 约束丢失,约束审计,constraint audit,governance decay, 压缩丢掉了什么,压缩摘要对照,compaction diff, OpenClaw trajectory,轨迹分析,run-based。 NOT for: 模型参数量估算、API 计费账单分析、Prompt 写作优化。" Observed in sandbox: no network, touches nothing outside the project, writes no memory.
Is [email protected] safe to install?
KillerSkills ran it in the KillerSkills honeypot sandbox (1 scenario) on 2026-10-26 and scored it 90/100 (Verified). It did what it claims and nothing else we could detect: no undeclared network egress, no credential reads, no persistent instruction writes. Summary: no network, touches nothing outside the project, writes no memory.
Behavior trace
Honeypot run · 0 s · 1 scenarios · 0 deviationsStatic findings
No findings across 1,346 lines in 6 files.