Word生成
- 作者仓库星标 435
- 作者更新于 实时读取
- 作者仓库 nvim
- 领域
- 文档
- 兼容 Agent
-
- Claude Code
- Cursor
- Cline
- Codex
- Windsurf
- Gemini CLI
- +20
- 信任分
- 88 / 100 · 社区维护
- 作者 / 版本 / 许可
- @benbrastmckie · 未声明 license
- Token 消耗评级
- 低消耗
- 接入复杂程度
- 需简单配置
- 是否需要外部 API Key
- 不需要
- 兼容的系统
- macOS
- 底层运行要求
- Python
- 文件与系统权限
-
- 只读
- 允许写入 / 修改
- Shell 执行
- 网络行为
- 仅限本地
- 安装命令数
- 26 条
档案由构建时根据 SKILL.md 与安装命令自动衍生,可能与作者实际意图存在差异。
需要注意: 未限定 allowed-tools,默认拥有全部工具权限。
---
name: skill-docx-edit
description: In-place DOCX editing routing to docx-edit-agent Thin wrapper that routes in-place DOCX editing…
category: 文档
runtime: Python
---
# skill-docx-edit 输出预览
## PART A: 任务判断
- 适用问题:PRD、RFC、README、项目说明或知识库整理。
- 输入要求:目标材料、限制条件、期望输出和验收方式。
- 证据边界:围绕“Context Pointers / Trigger Conditions / Direct Invocation”读取原文规则,不把推断写成作者承诺。
## PART B: 执行结果
- **01** 任务判断:确认你的需求是否属于PRD、RFC、README、项目说明或知识库整理,并标出输入、限制和预期结果。
- **02** 执行计划:优先按“Context Pointers / Trigger Conditions / Direct Invocation”拆成步骤,说明每一步会读取什么、修改什么、产出什么。
- **03** 交付结果:给出可复制的命令、文件改动、检查清单或内容草稿,并说明如何继续迭代。
- **04** 风险边界:结合 读取文件、写入/修改文件、执行终端命令、主要在本地完成、通常不需要额外 API Key 给出执行前确认项。
## Running Rules
- 读取文件、写入/修改文件、执行终端命令;主要在本地完成;通常不需要额外 API Key。
- 先小样例验证,再放大到真实任务。
- 交付时同时给结果、检查口径和下一步迭代建议。 原文出现了 `/edit`、`/convert`、`/scrape`、`/table` 这类斜杠命令;如果你的 Agent 支持命令触发,优先用命令开场,再补充目标和边界。
告诉 Agent 目标文件或材料、期望结果、不可改范围、是否允许联网或执行命令。本 Skill 的权限画像是:读取文件、写入/修改文件、执行终端命令。
先用一个小任务确认它会围绕“Context Pointers / Trigger Conditions / Direct Invocation”工作;涉及文件或命令时,先看 diff、日志、预览或测试结果。
检查最终产物是否包含明确结果、必要证据和下一步动作;如果输出泛泛而谈,就补充输入、边界和验收标准后重跑。
---
name: skill-docx-edit
description: In-place DOCX editing routing to docx-edit-agent Thin wrapper that routes in-place DOCX editing…
category: 文档
source: benbrastmckie/nvim
---
# skill-docx-edit
## 什么时候使用
- 把项目文档方向的常用动作沉淀成 Agent 可调用的技能 适合处理README、PRD、RFC、教程和知识库文档,核心价值是把输入、判断、执行、验证和交付边界固定下来,避免 Agent 泛泛回答。 把任务拆成可执行、可检查、可继续迭代的…
- 面向PRD、RFC、README、项目说明或知识库整理,优先处理能明确输入、步骤和验收标准的工作。
## 需要提供什么
- 目标材料、目录范围、期望结果和不可改动内容。
- 是否允许联网、执行命令、读写文件或调用外部服务。
## 执行规则
- 围绕「Context Pointers / Trigger Conditions / Direct Invocation」组织步骤,不把推断写成作者事实。
- 读取文件、写入/修改文件、执行终端命令;主要在本地完成;通常不需要额外 API Key。
- 先跑小样例,确认结果可检查后再扩大任务范围。
## 输出要求
- 给出最终产物、关键证据、验证方式和下一步动作。
- 信息不足时标记 unknown,不编造命令、平台或依赖。 作者原文负责流程事实;仓库文件负责来源和命令;流狐只补充适用场景、限制和质量判断。
skill "skill-docx-edit" {
输入层 -> 用户目标 + 目标文件 + 禁止范围 + 验收标准
上下文层 -> Context Pointers / Trigger Conditions / Direct Invocation
规则层 -> SKILL.md 触发条件 / 执行顺序 / 输出格式
运行层 -> Python | 读取文件、写入/修改文件、执行终端命令 | 主要在本地完成
安全层 -> 通常不需要额外 API Key + 小任务验证 + diff / 日志复核
输出层 -> 可复制结果 + 检查清单 + 下一步迭代
} DOCX Edit Skill
Thin wrapper that routes in-place DOCX editing requests to the docx-edit-agent, which edits documents using SuperDoc MCP with optional Word AppleScript integration for zero-friction workflow.
Context Pointers
Reference (do not load eagerly):
- Path:
.claude/context/formats/subagent-return.md - Purpose: Return validation
- Load at: Subagent execution only
Note: This skill is a thin wrapper. Context is loaded by the delegated agent, not this skill.
Trigger Conditions
This skill activates when:
Direct Invocation
- User explicitly runs
/editcommand with a .docx file - User requests document editing in conversation
Implicit Invocation (during task implementation)
When an implementing agent encounters any of these patterns:
Plan step language patterns:
- "Edit [file].docx to [change]"
- "Replace [text] with [text] in [file].docx"
- "Update the Word document [file]"
- "Add tracked changes to [file].docx"
- "Modify [file].docx with [instruction]"
- "Create a new document [file].docx"
File extension detection:
- Target file has extension:
.docx - Operation type: edit, replace, update, modify, create
Task description keywords:
- "DOCX editing"
- "Word document editing"
- "tracked changes"
- "in-place document edit"
When NOT to Trigger
Do not invoke for:
- Non-DOCX files (.xlsx, .pptx, .pdf, etc.)
- Document conversion (use skill-filetypes for /convert)
- PDF annotation extraction (use skill-scrape for /scrape)
- Spreadsheet table extraction (use skill-spreadsheet for /table)
- Reading document content without editing (use markitdown or /convert)
Execution
1. Input Validation
Validate required inputs:
file_path- Must be provided; for edit/batch must exist; for create mode may not exist yetinstruction- Natural language editing instruction (required)mode- One of: "edit", "batch", "create" (inferred from file_path and flags)
# For edit mode: validate source exists and is .docx
if [ "$mode" = "edit" ]; then
if [ ! -f "$file_path" ]; then
return error "DOCX file not found: $file_path"
fi
if [[ "${file_path##*.}" != "docx" ]]; then
return error "Source must be a .docx file: $file_path"
fi
fi
# For batch mode: validate directory exists
if [ "$mode" = "batch" ]; then
if [ ! -d "$file_path" ]; then
return error "Directory not found: $file_path"
fi
fi
# For create mode: validate .docx extension
if [ "$mode" = "create" ]; then
if [[ "${file_path##*.}" != "docx" ]]; then
return error "New file must have .docx extension: $file_path"
fi
fi
2. Context Preparation
Prepare delegation context:
{
"file_path": "/absolute/path/to/document.docx",
"instruction": "Replace ACME Corp with NewCo Inc. using tracked changes",
"mode": "edit",
"metadata": {
"session_id": "sess_{timestamp}_{random}",
"delegation_depth": 1,
"delegation_path": ["orchestrator", "edit", "skill-docx-edit"]
}
}
3. Invoke Agent
CRITICAL: You MUST use the Agent tool to spawn the docx-edit agent.
Required Tool Invocation:
Tool: Agent (NOT Skill, NOT Plan)
Parameters:
- subagent_type: "docx-edit-agent"
- prompt: [Include file_path, instruction, mode, metadata]
- description: "Edit {file_path}: {instruction}"
DO NOT use Skill(docx-edit-agent) - this will FAIL.
Agents live in .claude/agents/ or extension agent directories, not .claude/skills/.
The Skill tool can only invoke skills from .claude/skills/.
The agent will:
- Check if Word has the file open (macOS only)
- Save partner's unsaved work if needed (macOS only)
- Perform SuperDoc editing operations
- Reload the document in Word if it was open (macOS only)
- Confirm edit results
- Return standardized JSON result
4. Return Validation
Validate return matches subagent-return.md schema:
- Status is one of: edited, created, partial, failed
- Summary is non-empty and <100 tokens
- Artifacts array present with output file path
- Metadata contains session_id, agent_type, delegation info
Status value rules:
edited- File successfully modified in placecreated- New file successfully createdpartial- Some edits applied, others failed (batch mode)failed- No edits applied, error occurred- NEVER use "completed" (triggers Claude stop behavior)
5. Return Propagation
Return validated result to caller without modification.
Return Format
See .claude/context/formats/subagent-return.md for full specification.
Expected successful return:
{
"status": "edited",
"summary": "Replaced 3 instances of 'ACME Corp' with 'NewCo Inc.' in contract.docx with tracked changes (author: Claude)",
"artifacts": [
{
"type": "implementation",
"path": "/absolute/path/to/contract.docx",
"summary": "Edited DOCX with tracked changes"
}
],
"metadata": {
"session_id": "sess_...",
"agent_type": "docx-edit-agent",
"delegation_depth": 2,
"delegation_path": ["orchestrator", "edit", "skill-docx-edit", "docx-edit-agent"],
"tool_used": "superdoc",
"changes_count": 3,
"tracked_changes": true,
"word_was_open": true,
"word_reloaded": true
},
"next_steps": "Review tracked changes in Word"
}
Error Handling
Input Validation Errors
Return immediately with failed status if file not found, not a .docx, or instruction is empty.
Agent Errors
Pass through the agent's error return verbatim.
Tool Not Available
Return failed status with installation instructions for SuperDoc MCP or python-docx.
MUST NOT
- Run the postflight step (git commit, status update) -- that is the command's responsibility
- Modify the return from docx-edit-agent before propagating
- Load context files eagerly -- only reference them when needed
- Use the Skill tool to invoke docx-edit-agent (use Agent tool instead)
先判断是否适合
作者设计意图
作者的方法与取舍
边界和复核