wiki-fold compresses the raw wiki/log.md of an Obsidian vault into a fold page. You pick a batch size (default sixteen entries, set by the exponent k=4) or an explicit range. The skill drafts the rollup from the entries only, previews it, and applies it only when you say so.
Every child entry gets a key, a row in a table and an extractive one-sentence summary. Numbers and themes must trace back to the entries they came from.
What You Get
- A dry-run fold page with its ID, child range and read budget
- A child entries table matching the keys in the page frontmatter
- Key outcomes and themes that each cite their source entries
- One transaction that writes the fold page, an index entry and a log entry
- An operation ID and exact changed paths after apply
Common Use Cases
- Compressing a long operation log into a short monthly summary
- Reviewing what changed in the vault across a batch of operations
- Spotting contradictions between log entries and their pages
What Makes It Unique
The fold is extractive and structurally idempotent: it says ambiguous in source or source missing in place of guessing, and checks that entry counts, keys and table rows match. Child pages and logged entries stay untouched, and the hot cache is left alone.
Under The Hood
The skill is instructions plus a fold template reference. It needs no network access. For an explicit apply request it runs the plugin’s Python core script through the shell to inspect and apply one transaction bundle, which writes the fold page, the index and the log after you approve the plan hash. A Git checkpoint is a separate optional command. Helper agents may check entries and return extracts, but only the orchestrator assembles and applies the fold.