Crm Next Action

为机会池或客户列表逐条生成下一步动作、跟进理由和不推进原因。；use for crm, next-action, sales workflows；do not use for 伪造客户互动记录, 擅自发消息.

Security Scan

The skill is internally consistent with its CRM “next action” purpose: it is instruction-first, requires only python3, and the included script produces local, read-only structured reports without network calls or secret-requiring env vars.

Purpose & Capability

Name/description (generate next actions for opportunities) matches the code and resources: the script and templates produce structured briefs from input text/CSV. The script also implements additional audit modes (directory_audit, pattern_audit, csv_audit, skill_audit) that extend beyond a minimal 'next-action' generator; these are plausible utilities for reviewing input material but are broader than the short description.

Instruction Scope

SKILL.md stays within intended boundaries (generate suggestions, do not write back to CRMs, list 'do not fake interactions'). It instructs optionally running the bundled scripts. The script reads local files/directories and CSVs — if given arbitrary filesystem paths it can enumerate and read many text file types. The agent's runtime instructions do not direct network exfiltration or writing to external services.

Install Mechanism

There is no install spec; the skill is instruction-only with an included Python script. The only required binary is python3 and there are no downloads or third‑party package installs.

Credentials

The skill declares no environment variables, no credentials, and no config paths. It does read local files only when invoked with file/directory inputs — legitimate for its purpose but the user should avoid passing sensitive system paths.

Persistence & Privilege

always is false and the skill does not request persistent platform privileges or modify other skills. The script can write an output file if requested, but SKILL.md emphasizes read-only / dry-run modes and not writing back to external systems.

Assessment

This skill appears to do what it says: generate structured 'next action' suggestions and local audit reports. Before running the included script, review or run it in a sandbox and avoid passing sensitive directories (e.g., /, your home, or locations with secrets). Use the --dry-run flag and inspect outputs; do not pass production CRM credentials or paths that would allow the script to read unrelated confidential files. If you plan to let an agent invoke this autonomously, ensure it is only given appropriate input files and not broad filesystem paths.

CRM 下一步动作生成器

你是什么

你是“CRM 下一步动作生成器”这个独立 Skill，负责：为机会池或客户列表逐条生成下一步动作、跟进理由和不推进原因。

Routing

适合使用的情况

• 为这批机会生成下一步动作
• 同时给出不推进理由
• 输入通常包含：机会列表、阶段、最近互动
• 优先产出：机会摘要、下一步动作、优先级

不适合使用的情况

• 不要伪造客户互动记录
• 不要擅自发消息
• 如果用户想直接执行外部系统写入、发送、删除、发布、变更配置，先明确边界，再只给审阅版内容或 dry-run 方案。

工作规则

1. 先把用户提供的信息重组成任务书，再输出结构化结果。
2. 缺信息时，优先显式列出“待确认项”，而不是直接编造。
3. 默认先给“可审阅草案”，再给“可执行清单”。
4. 遇到高风险、隐私、权限或合规问题，必须加上边界说明。
5. 如运行环境允许 shell / exec，可使用：
   • python3 "{baseDir}/scripts/run.py" --input <输入文件> --output <输出文件>
6. 如当前环境不能执行脚本，仍要基于 {baseDir}/resources/template.md 与 {baseDir}/resources/spec.json 的结构直接产出文本。

标准输出结构

请尽量按以下结构组织结果：

• 机会摘要
• 下一步动作
• 行动理由
• 不推进原因
• 风险与阻塞
• 优先级

本地资源

• 规范文件：{baseDir}/resources/spec.json
• 输出模板：{baseDir}/resources/template.md
• 示例输入输出：{baseDir}/examples/
• 冒烟测试：{baseDir}/tests/smoke-test.md

安全边界

• 只生成建议，不直接写回 CRM。
• 默认只读、可审计、可回滚。
• 不执行高风险命令，不隐藏依赖，不伪造事实或结果。