Skill

CamScanner Skill is a document-processing capability package for agents. After installation, an agent can interpret natural-language requests and select and combine scan and image enhancement, text recognition, format conversion, PDF processing, and cloud saving and document management capabilities.

The Skill does not replace your agent. It teaches the agent when to use CamScanner, which command to choose, and which file and safety rules to follow.

When to use it

• Users should be able to state a goal while the agent plans the document-processing steps.
• A task combines multiple capabilities, such as enhancing a scan, recognizing an invoice, collecting results in Excel, and saving to the cloud.
• Your agent client or development platform supports Skills.
• You need richer intent routing, parameter selection, and error-handling guidance than a single command provides.

Use MCP when you only need standardized tools, or CLI when a script or terminal should execute explicit commands.

Quick setup

Ask an agent to install it

Copy this prompt to your current agent:

Help me install CamScanner Skill. Download and extract the latest package from https://data.camscanner.com/camscanner-cli/releases/latest/camscanner-skill-latest.zip. First confirm whether this agent supports Skills and identify its actual Skill directory. Read the package instructions, preserve existing configuration, and do not overwrite other Skills. When installation is complete, verify that the Skill is available and tell me which files changed and whether a restart is required.

The agent should confirm Skill support and locate the correct installation directory before copying files.

Manual installation

1. Download and extract

Download the latest Skill package.

The extracted package should contain:

camscanner/
├── SKILL.md
├── references/
└── scripts/

Keep the directory structure intact. SKILL.md is the entry point, references/ contains capability details, and scripts/ installs and upgrades the CLI.

2. Place it in the Skill directory

Move the complete extracted directory into the Skill installation directory used by your agent client. Locations and import methods differ by client, so follow that client's documentation.

Before installation:

1. Confirm that the client actually loads the target directory.
2. Back up the old directory if an earlier version is installed.
3. Do not delete or overwrite sibling Skills.

3. Install the bundled CLI

Run the appropriate setup script from the Skill directory:

macOS / Linux

bash scripts/setup.sh

Windows

powershell -ExecutionPolicy Bypass -File scripts/setup.ps1

The script detects the operating system and CPU architecture, downloads the matching camscanner-cli binary, and verifies the package. Current packages support macOS, Linux, and Windows on x64 and arm64.

On macOS and Linux, the default installation directory is ~/.local/bin. If it is not in PATH, follow the setup script instructions, then reopen the terminal or refresh the environment.

4. Sign in to CamScanner

Check authentication status:

camscanner-cli auth status

If you are signed out or the session has expired:

camscanner-cli auth login

The command opens a browser and remains in the foreground until authorization completes. Do not run it in the background. Run camscanner-cli auth status again afterward.

5. Reload the agent

Some clients discover a new Skill automatically; others require a Skill-list refresh or restart. After reloading, start document tasks with natural language.

Verify the integration

First verify the CLI:

camscanner-cli --version
camscanner-cli auth status

Then give the agent a clear, low-risk task:

Use CamScanner to recognize the text in this test image. Return the text only and do not save anything to the cloud.

A successful integration should let the agent:

1. Identify the request as OCR and select CamScanner Skill.
2. Confirm the input and expected output before execution.
3. Run the appropriate command instead of merely describing it.
4. Return the recognized content or clearly report the result location.

Supported capabilities

Category	Typical tasks
Scan and image enhancement	Brightness, sharpening, black-and-white and grayscale, shadow, dot-pattern and moiré removal, handwriting erasure, upscaling, and old-photo restoration
Text recognition and structured extraction	Single-image OCR, formula extraction, and structured invoice and receipt recognition
Document format conversion	Image to Word, Excel, Markdown, or PDF; PDF to Word, Excel, TXT, Markdown, or JPG; TXT to Word
Multi-page and batch processing	Merge up to 100 ordered images into PDF, Word, Excel, TXT, or Markdown
Translation and content editing	Layout-preserving translation, text replacement, and region deletion or movement after layout analysis
Watermarks and authenticity detection	Add or remove image and PDF watermarks; detect tampering and AI generation
Cloud saving and document management	Save results, combine search filters, download and move documents, and view folders

This site's capability scope is the intersection of Skill and MCP. For Skill parameters and limits, use the SKILL.md and references/ files in the installed package.

Usage examples

After installation, describe the goal directly:

Merge these three scans into a PDF in filename order and save it both locally and to my CamScanner cloud space.

Convert this contract PDF to an editable Word file named "Purchase contract-editable.docx".

Recognize this invoice and return structured data. Do not save it to the cloud.

Search my cloud space for the latest "project contract." Show candidates first and wait for confirmation before downloading.

Include the input, target format, save location, and cloud-saving preference when possible.

Update the Skill

Run the upgrade script from the installed Skill directory. It updates both the CLI and Skill files and exits when no update is available.

macOS / Linux

bash scripts/upgrade.sh

Windows / Node.js

node scripts/upgrade.cjs

If an upgrade fails, the script attempts to roll back. You can also run:

bash scripts/upgrade.sh --rollback

Troubleshooting

The agent cannot find the Skill

Confirm that the complete directory is in the location actually loaded by the client and that the entry file is still named SKILL.md. Refresh the Skill list or restart the client.

camscanner-cli: command not found

On macOS or Linux, add ~/.local/bin to PATH. On Windows, confirm that %LOCALAPPDATA%\camscanner-cli\camscanner-cli.exe exists and reopen the terminal.

Authentication still fails after sign-in

Run camscanner-cli auth status again. If the session expired, use camscanner-cli auth login to reauthorize in the browser.

The agent selected the wrong operation

State the input type, quantity, target format, and save location. For example, use “Merge three JPG images into one PDF” instead of “Process these files.”

Security and limits

• Input files are uploaded to CamScanner servers for processing. With -s, results are saved to the signed-in account.
• Credentials are managed by the CLI. Never ask an agent to display, copy, or log a plaintext token.
• Uploaded files should not exceed 40 MB. Images support JPG, JPEG, and PNG; documents support PDF, TXT, and Markdown.
• Password-protected PDFs are not supported.
• CLI -o may overwrite an existing file. Check the target path or choose a new filename.
• Confirm the file list and page order for multi-file tasks; do not rely on unordered wildcards.

Next steps

• See MCP for a standard tool connection.
• See CLI for command structure and scripting.
• See Supported capabilities for the available document capabilities.