- SIZE
- ARCH
- DIGEST
- HUMAN
Short answer
How can I audit my Docker image tags and metadata using AI?
Your AI uses the Docker Hub MCP to pull specific details like image size, architecture, and digests for any tag you specify. You can then ask your agent to compare versions or find outdated images. This breakdown shows you what you can do.
Audit outcomes
What you'll find
The AI pulls specific data points from your registry to help you make decisions.
SIZE
Image size analysis
The AI pulls the exact byte size for specific tags. You can identify bloated images that need pruning.
ARCH
Architecture checks
Your agent identifies the CPU architecture for any tag. This ensures you aren't deploying ARM images to x86 nodes.
DIGEST
Integrity verification
The AI retrieves unique digests for every tag. Use this to confirm you are running the exact build you intended.
HUMAN
Manual cleanup
The AI identifies which tags are old or too large. You decide which images to actually delete from the registry.
The workflow
What your AI does when the request arrives.
The AI moves through your registry to find the specific data you need.
Locate the repository
The AI finds the specific Docker Hub repository you want to inspect.
get_repositoryList available tags
The agent scans the repository to see every tag currently available.
list_tagsFetch tag details
The AI pulls the specific metadata, size, and digest for a single tag.
get_tagCompare and report
The agent compares the metadata between different tags to find discrepancies.
get_tag
Try it
Copy these to start.
Direct prompts to get your audit running.
Starting points
Replace the repository and tag names with your own. Use your specific image names.
Docker Hub Connector
You're all set. Choose your MCP client and follow the setup instructions.
https://edge.vinkius.com/vk_preview_ffdmDWTXZHfe3wwCZhc1wGAayNfR7ufnlBz3HHzP/mcpClaude Desktop
Follow the steps below to connect in seconds.
- 1In Claude Desktop, open Settings → Connectors.
- 2Click “Add custom connector” and paste the connector link above as the remote MCP server URL.
- 3Click Add and start a new chat — Docker Hub capabilities are ready to use.
{
"mcpServers": {
"docker-hub-mcp": {
"url": "https://edge.vinkius.com/vk_preview_ffdmDWTXZHfe3wwCZhc1wGAayNfR7ufnlBz3HHzP/mcp"
}
}
}List all the tags in the library/ubuntu repository.
What is the size and architecture of the library/nginx:latest tag?
Compare the digests for the my-app:v1 and my-app:v2 tags.
Find all tags in the my-service/api repository that are larger than 500MB.
Claude
ChatGPT
Cursor
VS Code
Windsurf
Claude Code
JetBrains
Cline
Start here
Connect Docker Hub once, then ask.
Follow the link to connect your account. Your credentials stay encrypted and the setup is handled on our end.
Connect Docker Hub to your AIFAQ
How this task behaves
- 01
Can my AI delete images from Docker Hub?
No. The AI can only read metadata and list tags. You must perform any deletions manually in the Docker Hub UI.
- 02
Can the AI push new images to my registry?
No. This MCP is read-only. It can inspect what is already there, but it cannot upload or modify images.
- 03
Does the AI see my private repository contents?
The AI can only see repositories that your connected Docker Hub account has permission to access.
- 04
Can I use this with any AI client?
Yes, as long as the client supports the Model Context Protocol (MCP).
- 05
Will the AI tell me which images are the oldest?
Yes, the AI can look at the tag data to determine which images were pushed most recently.
More questions about Docker Hub? The Connector page answers them. See everything the Docker Hub Connector can do
Connect Docker Hub to Claude, Cursor, ChatGPT & more
