Disclaimer
Cập nhật lần cuối:
What the generated documents are, what they are not, and why you must not treat them as a substitute for reading your code.
Tài liệu này do người vận hành trang soạn thảo và không phải là tư vấn pháp lý. Nếu bạn dựa vào nó cho một quyết định tuân thủ, hãy để một luật sư có chuyên môn tại khu vực pháp lý của bạn xem xét.
Tài liệu pháp lý này chỉ có bản tiếng Anh. Chúng tôi không cung cấp bản dịch, vì một chính sách bảo mật được dịch tự động có thể khiến bạn hiểu nhầm rằng nó đã được rà soát theo pháp luật của quốc gia bạn — và thực tế là chưa. Hãy dùng tính năng dịch của trình duyệt nếu cần; bản dịch đó là trách nhiệm của trình duyệt, không phải của chúng tôi.
1. Not professional advice
The Service produces technical documentation derived from source code. It is not legal, financial, security or engineering advice, and nothing it produces should be relied upon as such.
It is not a security audit. The presence or absence of a symbol in a generated document says nothing about whether your code is secure.
2. Accuracy and completeness
Generated documents are produced by deterministic static analysis. They are accurate about what they contain: file paths, symbol names, parameter lists, line ranges and import relationships are extracted from a parse tree.
They are necessarily incomplete, in ways you must understand before relying on them:
- Function and method bodies are not included unless you explicitly enable that option, and then only for a small number of files. Behaviour is therefore absent from the document.
- Files in languages without a loaded grammar are listed but not parsed. The document carries a coverage warning naming them.
- Files above the size limit, and projects above the file-count limit, are truncated. The document says so.
- Dynamic behaviour — dependency injection, reflection, runtime dispatch, string-built imports, configuration-driven routing — is invisible to static analysis and will be missing or wrong.
- Import resolution is heuristic. It does not read your bundler or TypeScript path configuration, and where an alias is ambiguous it declines to guess, leaving the edge unresolved.
- The document reflects your code at the moment you generated it. It is stale the instant you save a file.
3. Third-party AI systems
The Service is designed so that its output is provided to an AI system of your choosing. We are not affiliated with, endorsed by, or responsible for any AI provider.
Any conclusion, recommendation or code that an AI system produces from a generated document is that system's output, not ours. It may be wrong, may hallucinate symbols that do not exist, and may propose changes that break your system. You are responsible for reviewing and testing it before acting on it.
Providing a document to a third-party AI service is governed by that service's terms. Those terms may permit the provider to retain your submission and to use it to train models. Read them before you send anything, and do not send anything you would not send to that provider directly.
4. Confidential information
Generated documents can contain sensitive information derived from your code, including internal hostnames, credential-shaped string literals, proprietary architecture and personal data present in fixtures or migrations.
You are solely responsible for deciding whether to transmit a generated document to any third party. We provide no redaction guarantee. A secret-redaction feature is on our published roadmap but is not yet shipped. Until it is, treat every generated document as confidential to the same degree as the source it describes.
5. Trademarks
Third-party product names — including ChatGPT, Gemini, Claude, Qwen, DeepSeek, Grok, Copilot, VS Code, Cursor, JetBrains, Zed, Neovim, Claude Code, Codex, Aider and others — are the trademarks of their respective owners.
Their use on this site is nominative: to state, factually, that a plain Markdown file can be read by those products. No affiliation, endorsement or sponsorship is implied or should be inferred. We hold no integration with, licence from or relationship with any of them.
6. Open-source components
The Service is built on open-source software, including tree-sitter, web-tree-sitter, language grammars, fflate, and portions of Understand-Anything. Those components carry their own licences, principally MIT, and their own disclaimers, which apply to them and are not modified by anything here.
Attribution and licence texts are included in the application's repository and, where a licence requires it, in the distributed bundle.
7. Availability
The Service is provided free of charge and without any availability commitment. It may be modified, suspended or discontinued at any time, in whole or in part, without notice.
Because it holds none of your data, discontinuation costs you nothing but the tool. We nonetheless recommend that, if you depend on generated documents, you save them — they are yours, and they are ordinary files.