Long-form writing playbook
The monthly sweep found two accountabilities in this playbook that no longer match the role. It proposes trimming both and naming the escalation path.
The consent surface for an installed library of skills and tools: a four-segment review queue, the workspace inventory, a line-diff review, an assisted merge, bulk accept, and the refine room. Nothing here calls a model or writes anywhere — every action is a host port, wired in this demo to a note under the picker.
Each of these is its own route in a real install. The review surface carries four variants, because what the reader is consenting to changes what the surface is allowed to offer.
No action taken yet.
What wants your consent, and what your role-fillers changed while you were away.
The monthly sweep found two accountabilities in this playbook that no longer match the role. It proposes trimming both and naming the escalation path.
Adds a source-quality gate before the brief is handed on, after three briefs came back for the same missing step.
Wants to keep the speaker labels it strips today, so the analyst does not re-listen to work out who said what.
Upstream rewrote the matching rule in the same section you edited last month. The two versions disagree and need a merge.
A sealed listing: the publisher ships the body closed, so the review shows which regions changed rather than a line diff.
Keeps ordered lists intact when pasting from a document, instead of flattening them to plain paragraphs.
Splits the single readability pass into structure and line edits, so a long piece can be handed over halfway.
Turns a raw transcript into the decisions, the actions and who owns each one.
Walks a published page and reports links that no longer resolve.
Builds a question set from a research goal, ordered warm to sharp.
Scores a draft and points at the three sentences hurting it most.
Your edits were kept. The upstream change was woven around them, not over them.