Summary Budget Model And New Script

用于总结和更新预算数据模型及脚本技能。当用户发出【总结对话内容】指令时，从对话历史中提取新的脚本写法、数据模型和字段信息，更新到 platform-script 和 budget-data-model 技能中。

This skill will read conversation history and automatically append extracted scripts and data-model entries into other skills' files. Before installing, consider: 1) Do you want an automated process that can write into other skill directories? 2) Ensure it cannot accidentally copy secrets/PII from conversations—ask for a sanitization step (e.g., redact emails, API keys, passwords) before any write. 3) Require an explicit user confirmation prompt before making file changes, or restrict the skill to produce a proposed patch for manual review. 4) Verify and test in a safe/staging workspace to confirm paths and naming are correct (there are minor path/name inconsistencies in the docs). 5) Audit the update_log.md and backup directory regularly and limit write permissions so the skill can only modify intended files. If you need lower risk, prefer a variant that only suggests updates (read-only) rather than performing automatic writes.

Type: OpenClaw Skill Name: summary-budget-model-and-new-script Version: 1.0.0 The skill bundle is designed to automatically modify and update files in other skill directories (specifically `platform-script-skills` and `BudgetDataModelSkills`) based on conversation history. This cross-skill write capability, detailed in `SKILL.md` and `references/使用说明.md`, is a high-risk behavior that could be exploited via prompt injection to persist malicious code or corrupt data models across the agent's environment. While the intent appears to be productivity-focused knowledge management, the lack of robust sandboxing for these file operations makes it a significant security concern.