defuddle prepares web pages for an Obsidian wiki without fetching them. For an HTTPS URL it builds a plan that reports the host, the network access a cleaner would need and that nothing was executed. For a file you already have, it keeps the original bytes in the vault through a previewed capture.
The skill states why it does not run the Defuddle command line today: redirects and background requests cannot be held to the destinations you approved.
What You Get
- An inert plan for an HTTPS URL showing host, egress needs and execute false
- A status of available or configured for the optional cleaner, never relabelled as verified
- A previewed, hash-approved capture of a local source file into the vault
- A written explanation of why command line execution is held back
- A clean handoff point for the wiki-ingest skill
Common Use Cases
- Queueing a web article for later ingestion without fetching it
- Keeping a saved HTML or Markdown page as an immutable source in the vault
- Checking whether a Defuddle executable is present on the machine
What Makes It Unique
Most clipping helpers fetch and clean the page immediately. This skill refuses to run the upstream CLI until an adapter can enforce approved destinations after DNS and on each redirect, so it only plans URLs and retains local files. URLs with credentials, fragments, sensitive query parameters or non-public hosts are rejected.
Under The Hood
The skill is instructions only. It runs the plugin’s Python core script through the shell to build a URL plan, to verify the optional Defuddle capability, and to preview and apply a capture. The URL plan makes no network request. The capture copies a file you placed in the vault inbox into create-only raw storage after you approve the plan hash. It does not run the Defuddle program, install tools, or execute scripts in captured HTML.