confluence-cli

Note: All invocation methods below use the folder containing the current SKILL.md file as the workdir.

Script invocation method (must be executed directly, do not use uv run python or python）：

cd skills/confluence-cli && ./scripts/confluence_cli.py --json page get --page-id 3060336952

Incorrect example:

uv run python skills/confluence-cli/scripts/confluence_cli.py --json page get --page-id 3060336952
python skills/confluence-cli/scripts/confluence_cli.py --json page get --page-id 3060336952

1. Common subcommands (covering everyday scenarios)

• space
  • list [--start --limit --expand]
  • get --space-key [--expand]
• page
  • get --page-id [--body-format --expand --fresh]
  • by-title --space-key --title [--body-format --expand]
  • children --page-id [--start --limit --expand]
  • rename --page-id --title
  • publish-markdown --parent-id --title --markdown-path [--update-if-exists --body-format --expand]
• attachment
  • list --page-id [--start --limit --expand]
  • download --page-id [--output-dir --name --filter --all --start --limit --expand]
• search
  • --cql [--start --limit --body-format --expand]

1. Output format

• All invocations uniformly add --jsonafter the script and before the subcommand (example:./scripts/confluence_cli.py --json page get --page-id ...）

1. How to look up uncommon parameters/fields

• Run ./scripts/confluence_cli.py <command> --help to view the parameters of that command
• When deeper Confluence API fields are needed, you can extend the expand parameter in the script

1. Notes

• page get By default, keep normal cached reads; when you need to verify the body text right after updating a page, use --fresh to bypass the Confluence or proxy cache, for example:./scripts/confluence_cli.py --json page get --page-id ... --body-format storage --expand body.storage,version --fresh

1. Attachment download examples

• Download specified attachments (can be passed repeatedly --name）：./scripts/confluence_cli.py attachment download --page-id 3060336952 --output-dir ./attachments --name a.png --name b.png
• Download all attachments (automatic pagination):./scripts/confluence_cli.py attachment download --page-id 3060336952 --all --output-dir ./attachments
• Filtered download (regex):./scripts/confluence_cli.py attachment download --page-id 3060336952 --filter 'image2026-1-19_.*\\.png' --all --output-dir ./attachments

1. Publish Markdown examples

• Publish to a parent page (update if the same name exists):./scripts/confluence_cli.py --json page publish-markdown --parent-id 3061931928 --title "批量重置 Offset 功能测试" --markdown-path /path/to/doc.md
• Markdown tables support :---、:---:、---: such as left-aligned, centered, and right-aligned syntax, which will be converted to the table cell text-align style in Confluence storage.
• When publishing local images, a single Confluence size attribute is automatically generated based on the maximum display frame: the default maximum width is 1000, maximum height is 800, which can be overridden with --image-max-width / --image-max-height ; only write the ac:width or ac:heightthat triggers scaling, and the original attachment image is not modified.
• If you need to override the display size of a single image, you can use the Markdown title:![图](./a.png "confluence-width=1200")、![图](./a.png "confluence-height=600")、![图](./a.png "confluence-size=original")。

Resources

• confluence_cli.py: main CLI entry point, responsible for reading configuration and initiating API calls.
• confluence_api_client.py: SDK wrapper layer, which consolidates common API calls.