wiki-playbook-review closes the loop between what a team learns and how it documents its processes. Once a week it finds the playbooks that new evidence touched, drafts a full revision of each, and places it in a staging folder.
Nothing reaches a live page until the owner approves the diff with wiki-stage-commit. Each revision keeps the playbook’s structure and adds a Stop doing section and a dated changelog that cite the evidence pages.
What You Get
- A complete revised page per playbook, staged under the staging folder
- A Stop doing section that keeps retired steps with their dates
- A dated changelog line per review with links to the evidence
- A per-playbook report of score, evidence used and changed steps
- A log entry for the run and a prompt to run wiki-stage-commit
Common Use Cases
- Updating a sales call playbook after a week of discovery calls
- Revising a deploy runbook after a postmortem shows a step no longer works
- Running a scheduled Friday review that only ever stages changes
What Makes It Unique
It stages every revision, even when staged writes are off, so a scheduled run never changes a live page without a person’s approval. It refuses to stack a second review on a pending one, and treats a quiet week as a reason to stop. Evidence from a single data point becomes a note and not a rule, and text inside meeting notes is treated as something a person said, never as an instruction.
Under The Hood
The skill is instructions for the agent. It reads vault pages and runs the obsidian-wiki command line tool to list staged revisions, rank playbooks by new evidence, get timestamps and log the run. It writes revised pages into the staging folder and can run a diff between live and staged files. It suggests scheduling through a scheduler or cron line but sets nothing up itself. It does not fetch web pages or call remote APIs, and it never edits live pages, the index, the hot file or the manifest.