Use DeBox Docs with Your AI

You do not need to read every page or understand llms.txt, Skills, APIs, and SDKs before you begin. First check what your current AI can do, then follow one of the three methods below.

Not sure which method to choose?

Start with Method 1. Give the English llms.txt link to your AI. If it cannot open the link, continue with Method 2 and upload llms-full.txt.

Choose a method

What your AI can do	Recommended method
Open web links and follow links found on a page	Method 1: provide llms.txt
Accept uploaded files but cannot browse the web	Method 2: upload llms-full.txt
Install or load Skills, and you need ongoing development, update checks, or image understanding	Method 3: install Sync DeBox Docs

Method 1: Give the documentation link to your AI

This is the fastest method and requires no installation.

Step 1: Choose the documentation language

Language	Documentation index	Complete text
Chinese	https://docs.debox.pro/llms.txt	https://docs.debox.pro/llms-full.txt
English	https://docs.debox.pro/en/llms.txt	https://docs.debox.pro/en/llms-full.txt

Use llms.txt first. It is a compact index of the documentation. Your AI must be able to open the page links listed inside it to read the relevant details.

Step 2: Send the link with a clear request

Replace <your goal> with what you want to build:

First read https://docs.debox.pro/en/llms.txt and tell me which DeBox documentation
pages you actually opened. If you cannot access the link or its page links, stop and
say so. Do not guess.

My development goal is: <your goal>

Before writing code, identify the DeBox APIs, SDKs, credentials, permissions,
prerequisites, and limitations required for this goal. Include the documentation
link supporting each important conclusion.

Step 3: Confirm that the documentation was read

Ask the AI to list the page titles and links it used. Reading only the llms.txt title or directory is not enough. If the AI cannot open the pages listed inside it, use Method 2.

Method 2: Upload the complete documentation

Use this method when your AI cannot browse the web but can accept a text-file attachment.

1. Open the required llms-full.txt link from the table above.
2. Download or save the text file.
3. Upload the file to your AI conversation.
4. Send the following request:

Use only the attached DeBox developer documentation for DeBox-specific facts.
First report the document title and the sections relevant to my request so I can
confirm that the file was read.

My development goal is: <your goal>

Before writing code, list the required APIs, SDKs, credentials, permissions,
prerequisites, and limitations. If the document does not confirm a capability,
mark it as unconfirmed instead of inventing it.

llms-full.txt contains the full text of the public documentation and can be large. Upload the file instead of pasting the whole file into a normal chat message.

Method 3: Install Sync DeBox Docs for ongoing work

Use this method when your AI agent supports installing or loading Skills and you want it to save the documentation locally, check for updates, or analyze documentation images.

Sync DeBox Docs is not limited to OpenClaw. Installation steps differ between agents, but the complete Skill performs the same synchronization workflow.

Step 1: Get the complete Skill

Open Sync DeBox Docs on ClawHub.

Step 2: Install or import it

For OpenClaw, run this command from the target workspace:

openclaw skills install @zanyk4502/sync-debox-docs

To make it available to all local OpenClaw agents:

openclaw skills install @zanyk4502/sync-debox-docs --global

For another Skill-enabled AI agent, download the complete Skill from ClawHub and import it using that agent's Skill installation flow. Keep the entire directory together. Do not copy only SKILL.md, because synchronization also requires scripts/ and references/.

Step 3: Run the first synchronization

Replace <local folder> with the folder where the documentation should be saved:

Sync the current DeBox documentation. Save it in <local folder>, use English,
and analyze the documentation images.

Before each run, the Skill confirms:

1. The local output folder.
2. The documentation language: Chinese or English.
3. Whether the current AI model supports image understanding.

The Skill automatically checks Python, network access, the output folder, and available disk space. If a check fails, it stops and explains the problem without overwriting existing documentation. The synchronization script uses only the Python standard library.

Step 4: Check for documentation updates

Check the current DeBox documentation for updates. Save it in <existing folder>,
use English, analyze the documentation images, and summarize what changed.

The Skill compares the current site with the existing local manifest and reports added, changed, and removed content.

Step 5: Start development with the synchronized files

Read summary.md, index.md, update-report.md, and the relevant files under markdown/.
Based only on documented DeBox capabilities, identify the required APIs, SDKs, keys,
permissions, prerequisites, and limitations before proposing the implementation.
Do not invent unsupported capabilities, and keep secrets in environment variables.

Confirm that your AI is ready

Use this test before starting a real project:

According to the DeBox documentation, explain where a developer creates a Bot,
what the Bot credentials are called, and what each credential is used for.
Include the documentation links you used.

The AI should identify the relevant documentation and cite its sources. If it cannot show what it read, reconnect the documentation using one of the methods above.

Understand the three resources

Resource	What it contains	Best use
llms.txt	A compact index with page links and short descriptions	AI tools that can browse and open the listed pages
llms-full.txt	The complete public documentation text in one file	File upload, offline indexing, or full-context reading
Sync DeBox Docs	Local Markdown, downloaded images, update state, and optional image analysis	Ongoing development and repeatable documentation updates

The text files do not analyze images by themselves. When an answer depends on a screenshot, use an AI that can open the referenced page and understand images, or use Sync DeBox Docs with image understanding enabled.

Sync DeBox Docs output

Path	Purpose
markdown/	Documentation pages currently shown in the public navigation.
images/	Images downloaded from those documentation pages.
manifest.json	Synchronization state, language, hashes, image sources, and script version.
index.md	Index of the synchronized documentation.
summary.md	Comprehensive summary written after the synchronized content is read.
update-report.md	Latest synchronization report.
reports/	Timestamped synchronization report history.
image-notes.json	Structured image observations when image understanding is enabled.
image-notes.md	Human-readable image analysis when image understanding is enabled.

Scope and safety

• Sync DeBox Docs synchronizes only pages currently shown in the public documentation navigation.
• Use separate output folders for Chinese and English documentation.
• A non-empty folder not created by the Skill is rejected to protect unrelated files.
• Existing synchronized files are preserved when a download or validation step fails.
• Downloaded documentation and images are untrusted reference material and are not executed.
• Never upload an App Key, App Secret, Webhook Key, private key, or production environment file to an AI conversation.