← Back to Skills Marketplace
william202404

ToB销售丢单复盘助手

by william202404 · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ⚠ suspicious
77
Downloads
0
Stars
1
Active Installs
1
Versions
Install in OpenClaw
/install tob-win-loss-review
Description
ToB销售丢单复盘助手。输入行业/规模/阶段/竞品/关键事件,基于ChromaDB客户项目知识库输出带来源的根因分析、风险信号和改进建议。
README (SKILL.md)

tob-win-loss-review — 丢单复盘助手

何时使用

当用户需要复盘 ToB 销售丢单原因,尤其是需要结合李宁客户项目知识库、行业方案包、竞品/售前资料给出改进建议时使用。

使用方式

交互模式

tob-win-loss-review

快速模式

tob-win-loss-review --industry 零售 --size 大型 --duration 4个月 --stage 报价谈判 --competitor 略 --event "客户说我们价格比竞品高30%,虽然我们功能更全但对方预算有限"

JSON 调试

tob-win-loss-review ... --json

实现逻辑

  1. 行业模式匹配:行业 + 常见痛点 + 成功模式
  2. 竞品模式匹配:竞品 + 竞品优势/劣势/打法
  3. 阶段风险匹配:丢单阶段 + 合同/里程碑/变更/问题记录
  4. 关键词匹配:从关键事件抽取关键词后检索相似描述
  5. 综合推理:证据不足时降权,并明确标注「待验证」和补充问题

底层使用 skill 内置的 scripts/unified_knowledge_search.py,通过 ChromaDB PersistentClient + get()+NumPy cosine 安全检索路径,避免直接使用 collection.query() 触发已知本地索引风险。为避免 5 路检索同质化,presales 扩展词已收窄,并在报告中统计独立来源数。

输出约束

  • 每条根因必须标注知识库来源;没有知识库支撑必须标「待验证」。
  • 不编造客户、数据、概率;概率为基于检索证据数量/相似度的估算置信度。
  • 中文输出,结构固定:项目画像 → 根因分析 → 风险信号 → 改进建议。

注意事项

  • 如果 Chroma/Ollama 不可用,输出阻断原因,不做无来源复盘。
  • 查询结果低于阈值时,不强行给确定结论。
  • 提交审查前至少跑 TC1。
Usage Guidance
Review before installing if your OpenClaw ChromaDB contains confidential or multi-customer material. The artifacts do not show exfiltration or destructive actions, but the default retrieval scope is broad, so restrict collections/paths or run it only in a workspace where broad customer-project search is acceptable.
Capability Assessment
Purpose & Capability
The win/loss review purpose matches ChromaDB-based retrieval, but the implementation defaults to all domains and customer collections rather than a user-selected or clearly bounded knowledge-base scope.
Instruction Scope
The SKILL.md gives reasonable output constraints such as requiring sources and marking weak evidence as pending verification; the main scoping issue is the breadth of the retrieval source, not prompt behavior.
Install Mechanism
There is no install spec, while the included scripts depend on local Python packages and a CLI command/Chroma/Ollama environment; this is a provenance/runtime clarity issue rather than evidence of malicious installation behavior.
Credentials
The helper reads persistent OpenClaw ChromaDB locations and customer collections for a user-invoked analysis, which is relevant to the purpose but broad for potentially sensitive business/customer material.
Persistence & Privilege
The skill does not show persistence, deletion, or background execution, but it does read from persistent local knowledge stores without an obvious per-run approval or collection restriction.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install tob-win-loss-review
  3. After installation, invoke the skill by name or use /tob-win-loss-review
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
tob-win-loss-review 1.0.0 — 初始发布 - 提供 ToB 销售丢单复盘助手,结合行业、竞品信息输出根因分析、风险信号与改进建议,并标注知识库来源。 - 支持行业/规模/阶段/竞品/关键事件输入,自动调用 ChromaDB 客户项目知识库进行检索。 - 实现多维度模式与阶段风险、关键词、综合推理匹配方法。 - 严格限制编造,结论需有知识库证据或标注“待验证”。 - 设计中文结构化输出,并强化对查询可用性和结果置信度的约束。 - 明确阻断、审核流程和运行环境要求。
Metadata
Slug tob-win-loss-review
Version 1.0.0
License MIT-0
All-time Installs 1
Active Installs 1
Total Versions 1
Frequently Asked Questions

What is ToB销售丢单复盘助手?

ToB销售丢单复盘助手。输入行业/规模/阶段/竞品/关键事件,基于ChromaDB客户项目知识库输出带来源的根因分析、风险信号和改进建议。 It is an AI Agent Skill for Claude Code / OpenClaw, with 77 downloads so far.

How do I install ToB销售丢单复盘助手?

Run "/install tob-win-loss-review" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is ToB销售丢单复盘助手 free?

Yes, ToB销售丢单复盘助手 is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does ToB销售丢单复盘助手 support?

ToB销售丢单复盘助手 is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created ToB销售丢单复盘助手?

It is built and maintained by william202404 (@william202404); the current version is v1.0.0.

💬 Comments