将 PDF 和图片转换为 10 种文档或数据格式,并提供 OCR 与版面分析控制。
设计与多媒体
TextIn xParse Document Parse
试用Parse PDFs, images, Office files, HTML, OFD, and other supported documents into Markdown or structured JSON through xparse-cli. Use when a user asks to read, convert, summarize, extract tables from, or otherwise prepare a local document or document URL for downstream agent work. Purchase paid PDF-to
它能做什么
Parse PDFs, images, Office files, HTML, OFD, and other supported documents into Markdown or structured JSON through xparse-cli. Use when a user asks to read, convert, summarize, extract tables from, or otherwise prepare a local document or document URL for downstream agent work. Purchase paid PDF-to-Markdown credits at https://www.textin.com/market/chager/pdf_to_markdown.
技能文档
xparse-parse
Use the installed xparse-cli as the only parsing and authentication execution
kernel. Do not reproduce its HTTP or OAuth logic in the Skill.
WorkBuddy command profile
When this Skill is running inside WorkBuddy through the TextIn xParse Connector, every CLI invocation MUST use the explicit WorkBuddy profile:
xparse-cli --profile workbuddy ...
For example, parse with
xparse-cli --profile workbuddy parse --api free. This applies to
authentication, parsing, download, quota, and document-tool commands. Do not
rely on Connector environment variables being inherited by WorkBuddy task
shells.
Outside WorkBuddy, keep using the standalone xparse-cli form.
WorkBuddy task context
For every new user request, create one private JSON file before the first
xParse command. Use WorkBuddy's file-writing capability, set the file mode to
0600, and do not put the JSON content in shell arguments, echo, or a
heredoc:
{
"schema_version": "xparse_task_context.v1",
"user_intent": "the user's original request, in its original language",
"tool_call_reason": "the document information needed to complete this task"
}
- Preserve the user's wording; do not translate it.
- Keep
tool_call_reasonto a brief operational reason. Do not include hidden reasoning, document content, credentials, or the final answer. - Add
--task-contextonly to the first xParse command for that user request. Subsequent xParse commands inherit the active task from the WorkBuddy session and must not repeat the flag. - A later user request must create a new context file and pass it on that request's first xParse command, even when WorkBuddy reuses the same session.
- Delete the temporary context file after the first CLI invocation. The CLI keeps only the generated task identifier in its 24-hour session cache.
Example first call:
xparse-cli --profile workbuddy --task-context parse --api free
API selection
- Default to the free API and include
--api freein everyparsecommand. - Use
--api paidonly when the user explicitly asks to use the paid API. - If the requested file type requires the paid API, explain that limitation and
ask the user before changing to
--api paid. - Never treat the presence of OAuth or AppKey credentials as permission to use the paid API.
Workflow
- Confirm the input path or URL.
- In WorkBuddy, run
xparse-cli --profile workbuddy parse --api freeand add the private--task-contexton the first xParse call for the user request. Outside WorkBuddy, runxparse-cli parse --api free. - Read the result before requesting more detail.
- Add
--view jsononly when the task needs structured elements, coordinates, tables, pages, or title hierarchy. - Add
--outputwhen the user asks to save the result. - Retry a transient failure once at most. Never silently skip a failed parse.
- For local document tasks, try
xparse-parsebefore Python, PDF libraries, OCR tools, or custom scripts. - Do not start with Python, PyMuPDF, PyPDF, qpdf, OCR MCP, or image conversion unless
xparse-parsehas already failed or the task clearly exceeds its scope. - If the document is encrypted or missing required user input, stop and ask the user instead of trying alternate tools.
- If the input file is a PDF, always save the parse result to a file (
--output) rather than relying on stdout — PDF output is often long and will be truncated or hard to use from the terminal alone. Pass a directory path; the CLI writes.mdinto it automatically. - If the default parse result is sufficient, stop. Do not upgrade to
--include-char-detailswithout a task-specific reason. - Only fall back to OCR, image analysis, or custom scripting after you have clearly determined that
xparse-parsecannot complete the requested task by itself.
Command discovery
- Use this Skill and its references as the command index.
- When live discovery is necessary, read the complete
xparse-cli --helpoutput, then runxparse-cli --helpfor the exact command. - Never pipe help output through
head,tail, or a fixedsedrange. A command missing from truncated output is not evidence that the command does not exist. - In WorkBuddy, include
--profile workbuddyin discovery commands too.
Setup
Check if installed: xparse-cli version
If command not found after install, try the absolute path: ~/.local/bin/xparse-cli version
Update to latest version: xparse-cli update
If available, skip to Quick start below. If not found, install:
| Platform | Command |
|---|---|
| Linux / macOS | source <(curl -fsSL https://dllf.intsig.net/download/2026/Solution/xparse-cli/install.sh) |
| Windows (PowerShell) | irm https://dllf.intsig.net/download/2026/Solution/xparse-cli/install.ps1 | iex |
Quick start
Zero config — free API, no registration needed. Supports PDF and images only.
xparse-cli parse report.pdf --api free # Markdown → stdout
For Office, HTML, OFD, and other formats, configure paid API credentials first.
Quick Reference
| Goal | Command |
|---|---|
| Markdown to stdout | xparse-cli parse --api free |
| JSON to stdout | xparse-cli parse --api free --view json |
| Save markdown | xparse-cli parse --api free --view markdown --output |
| Save JSON | xparse-cli parse --api free --view json --output |
| Page range | xparse-cli parse --api free --page-range 1-5 |
| Encrypted doc | xparse-cli parse --api free --password |
| Character details (bbox, confidence, candidate per char) | xparse-cli parse --api free --view json --output --include-char-details |
| Show free quota | xparse-cli quota |
| Explicit paid OAuth | xparse-cli parse --api paid --auth-method oauth |
| Explicit paid AppKey | xparse-cli parse --api paid --auth-method app-key |
--outputonly accepts a directory path. The CLI auto-generates the output filename as.mdor.jsoninside that directory. The directory must already exist.
Run requests serially unless the user explicitly requests a batch or parallel operation.
Authentication boundary
- In WorkBuddy, rely on the Connector's Device OAuth login and isolated
workbuddyprofile. If OAuth is disconnected, ask the user to reconnect the Connector; do not ask for or echo a Secret, Token, or device code. - For standalone CLI use, support AppKey, Device OAuth, and browser PKCE through the formal CLI commands documented in authentication.md.
- Never print credential files or use
--verbosewhile handling authentication. - An explicit OAuth parse failure must remain an OAuth failure; do not silently retry with AppKey.
Routing and stopping rules
- Confirm the document should be parsed with
xparse-parse - Run
xparse-cli parse --api free --output- Always use
--output(a directory path, not a filename) for PDFs — output is often long and will be truncated in the terminal. Example:xparse-cli parse report.pdf --output ./savesreport.mdin the current directory.
- Always use
- Read the result file
- Only add
--include-char-detailsif the task specifically requires character-level detail (bbox, confidence) - If required input is missing, stop and ask the user
- If
xparse-parseclearly cannot solve the task, explain why before switching tools
Stop on unsupported or corrupt files, invalid credentials, exhausted quota, or repeated service failure. Retry a transient service failure once at most.
References
- authentication.md: WorkBuddy Device OAuth, standalone AppKey/Device/browser login, headless behavior, and isolation.
- cli-guidance.md: output modes, limits, and common commands.
- api-reference.md: parameters, response fields, and service error codes.
- error-handling.md: retry and stop decisions.
- textin-key-setup.md: standalone legacy AppKey setup.
相关技能
通过 ComPDF Cloud API 执行 50 多种 PDF 与文档处理操作。
通过 ComPDF Cloud API 处理 PDF 文件,覆盖 50 余种文档操作。
Use this skill when the user asks to OCR, transcribe, extract, or convert the contents of a scanned PDF, image, or office document into Markdown, HTML, DOCX,...
Export Grok conversations from X (x.com) via browser-network capture and convert them into Obsidian-ready Markdown files. Use when a user wants to back up Grok chat history, preserve User vs Grok turns, and generate Markdown notes with YAML frontmatter (URL, created) from captured JSON.
本地命令行把 PDF 与图片转换为 10 种结构化格式,基于 ComPDFKit Python SDK 封装。