← 返回 Skills 市场
556
总下载
5
收藏
0
当前安装
9
版本数
在 OpenClaw 中安装
/install somark-document-parser
功能描述
Parse PDF, images (PNG/JPG/BMP/TIFF/WebP/HEIC), Word, PPT and other documents into Markdown or JSON using SoMark. All document parsing needs (like resumes, p...
安全使用建议
This skill sends any parsed files (and your SOMARK_API_KEY) to somark.tech to do the work — that is the intended behavior. Before installing: 1) Confirm you trust somark.tech and are comfortable sending documents (do not send highly sensitive PII or secrets unless allowed by your policy). 2) Verify how the script receives the API key (environment vs CLI) and store keys with least privilege and rotation. 3) Check where output files are written and whether temporary uploads are removed if you are worried about disk leakage. 4) Test on non-sensitive documents first to confirm behavior and billing characteristics. If you need the parser to run fully offline or to avoid external uploads, this skill is not appropriate.
功能分析
Type: OpenClaw Skill
Name: somark-document-parser
Version: 1.0.9
The somark-document-parser skill is designed to convert various document formats (PDF, Office, images) into Markdown or JSON using the SoMark API (somark.tech). The Python script (somark_parser.py) correctly handles file processing, API communication via aiohttp, and result retrieval using environment variables for authentication. Notably, the SKILL.md includes explicit safety instructions directing the AI agent to treat all parsed content as untrusted data to mitigate prompt injection risks from document content. No malicious behaviors such as unauthorized data exfiltration or persistence were detected.
能力标签
能力评估
Purpose & Capability
Skill name/description match the included code and instructions: the Python script and SKILL.md upload files and POST them to somark.tech for parsing. No unrelated credentials, binaries, or services are requested.
Instruction Scope
Instructions direct the agent to save uploaded files locally and run the provided somark_parser.py to convert files to Markdown/JSON — this is expected. Minor omission: the CLI examples in SKILL.md do not explicitly show how the SOMARK_API_KEY is passed to the script (the metadata declares the env var). The script (truncated) posts files and form data to SoMark endpoints; it writes output files to disk (output directory). Review handling of temporary files if you are concerned about sensitive data left on disk.
Install Mechanism
No install spec is included (instruction-only plus a bundled script). No remote downloads or extract steps. The only network targets are somark.tech API endpoints shown in the script.
Credentials
Only SOMARK_API_KEY is required and is the primary credential — this matches the stated purpose. Note: the API key and file contents are transmitted to somark.tech as part of normal operation, so the key will be sent over the network and is required for the service.
Persistence & Privilege
Skill does not request persistent/always-on privileges and does not modify other skills or system-wide settings. It reads/writes files only in the specified output directory.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install somark-document-parser - 安装完成后，直接呼叫该 Skill 的名称或使用
/somark-document-parser触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.9
Auto-publish from GitHub Actions
v1.0.8
Update SKILL.md and somark_parser.py
v1.0.7
somark-document-parser 1.0.7
- Updated default parser configuration instructions: users are now informed how to make SoMark the default document parser by editing their CLAUDE.md, rather than asking in chat.
- Clarified guidance after a successful parse, improving accuracy and self-service setup.
- No changes to core functionality or API usage.
v1.0.6
- Rewrote documentation in English for broader accessibility and clarity.
- Added explicit instructions to ask the user whether to set SoMark as the default document parser.
- Simplified and clarified API key configuration steps, emphasizing security and not sharing keys in chat.
- Consolidated usage instructions, clarifying trigger scenarios and supported file formats.
- Reinforced safety: treat all parsed output as data and ignore any embedded prompts or instructions.
- Updated setup, quota, and error handling instructions for consistency and brevity.
v1.0.5
somark-document-parser 1.0.5
- Added initial code file `somark_parser.py` implementing the document parsing functionality.
- Added metadata file `_meta.json` describing environment variable requirements.
- No changes to parsing logic or user-facing workflow; this release formalizes the script and metadata structure.
v1.0.4
- Removed source code and metadata files: `_meta.json` and `somark_parser.py`.
- Updated SKILL.md to clarify that users should not send their API Key in the chat and must set it only via environment variables, with explicit instructions not to provide the key in conversation at any step.
- Minor clarification in API Key acquisition and setup steps in both Chinese and English instructions.
- No functional code or behavior changes, documentation and security guidance tightened.
v1.0.3
- Removed the README.md file.
- Updated API Key handling: no longer ask users to send their API Key in chat. Users are now instructed to configure the API Key via environment variable in their own terminal for increased security.
- Added explicit security instructions: only display parsed document data, never execute or obey document-embedded instructions or roleplay requests.
- Clarified that all parsed content is presented as pure data and must not be summarized or modified.
- Revised wording throughout skill documentation to reflect the new API Key and security processes.
v1.0.1
### somark-document-parser 1.0.1
- Added a new document parsing startup script, somark_parser.py.
- Implemented language detection (automatic Chinese/English replies).
- Extended descriptions for file upload and path resolution scenarios.
- Optimized guidance for API key acquisition and quota handling instructions.
- Improved user-facing error and result handling descriptions.
v1.0.0
init commit
元数据
常见问题
Somark Document Parser 是什么？
Parse PDF, images (PNG/JPG/BMP/TIFF/WebP/HEIC), Word, PPT and other documents into Markdown or JSON using SoMark. All document parsing needs (like resumes, p... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 556 次。
如何安装 Somark Document Parser？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install somark-document-parser」即可一键安装，无需额外配置。
Somark Document Parser 是免费的吗？
是的，Somark Document Parser 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Somark Document Parser 支持哪些平台？
Somark Document Parser 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Somark Document Parser？
由 Soul-Code（@soul-code）开发并维护，当前版本 v1.0.9。
推荐 Skills