Editorial flows.

Automation should not wait for a person, and a person should still be able to steer it. This recipe covers Review Requests, the Notification Center, and the way an editor's steering reaches the next run. The examples are Shipped and Distinctive Mark, both open-code demonstration sites.

Flag, soft review or hard gate.

A step picks how much it needs a person. A flag waits for nothing. A soft review waits until a deadline, then proceeds and records "proceeded without review". A hard gate never proceeds without an answer.

papyrus reviews raise --subject-type story --subject-id ID \
  --mode soft_review --reason "Two drafts to read"

Where it is used

• Cited Trends raises one soft review per weekly run.
• A Stage Bot that holds or fails raises a flag.
• A bot chain that reaches its hop limit raises a hard gate.

Ignore what you like.

The console home lists the open requests. Editors answer some, ignore others, and the system keeps running. Only a hard gate waits. Marks and words accompany every colour.

papyrus reviews list --state open

Safe to ignore

If you only look, nothing changes. See Review Requests.

Say what you want next time.

An answer can approve, reject, or give feedback with a steer such as "adjust" or "not this kind". The step that raised the request reads it at the start of its next run, and the console then shows "Applied in run".

papyrus reviews steering --raised-by idea-pipeline

Cited Trends

A declined draft is not proposed again, and notes become feedback for every draft in that batch. See Cited Trends.

How it stays honest.

1. Review is never a story state. Stories move on their own stages.
2. Timeouts are recorded. "Proceeded without review" is written down.
3. Steering is applied visibly, with the run that used it.
4. Recipients are configuration, not code.

Related: Automations, Stage Bots.