[titerman]

This PR builds on #825: its evaluations and rule decisions framed this work. It adds rules for the problems that kept coming back in Claude's docs drafts over the last two weeks, across different articles and often after an earlier correction.

Why these changes

Each change starts from a recurring correction, compared with editor and writing-planner as they stand after #825. I reviewed every change individually: many were rewritten or narrowed, and the ones that didn't hold up as general rules were dropped.

The main finding: most rules exist, but their tests or examples don't match the cases that actually come up. As a result, most changes extend an existing line with a sharper test or a ❌/✅ pair taken from accepted wording, instead of adding a new rule.

Fit with #825

The changes follow #825's decisions and reverse none of them. Corrections that conflicted with a decision stayed out, and most changes extend an existing line rather than add to editor.

What changes

• Headings, in the writing-planner heading tree: each heading answers its own question, names what the reader wants to do, and differs from its siblings on what the reader decides by.
• Precision, in editor Area 4: terms explained before use, comparisons with a stated base, a test for second readings, and a reason for every constraint.
• Framing and order, in Area 12: purposes that name a real outcome, no implied fault, conditions before outcomes, causes before reactions.
• Claims, in Area 13: no under-scoping, and verbs and owners that are literally true.
• Editing habits, in the editor process sections: referent tests on every proposed rewrite, and a cut instead of a patch for a claim proven wrong.

Next

Test these rules the way #825 tested its own, and add the correction-to-rule loop to doc-author.