The context graph
A legal notice, a help article and a release note can describe the same product with different vocabulary and levels of detail. The appropriate wording depends on the audience, the surface and the date.
A content context records those conditions as coordinates and associates them with writing guidance. People, checks and AI assistants can retrieve the guidance that applies to the content they are working on.
One decision, five rules
A feature rename illustrates why a rule needs a scope:
| Where the words sit | What the rule is |
|---|---|
| Help articles | Use the new name |
| The API parameter | Must not change |
| The changelog | Keeps the old name, because that is what shipped |
| The migration guide | Needs both, side by side |
| Support replies | Recognise the old name for a year, never use it |
The rename requires different rules for different surfaces. A project-wide list of banned and preferred terms cannot express all of them. The graph associates each rule with the scope where it applies.
Profiles and coordinates
A context is a set of coordinates:
| Coordinate | The question it answers |
|---|---|
| Audience | Who is it for |
| Surface | Where does it appear |
| Register | In what register |
| Market | For which market |
| Validity | From when until when |
A profile groups coordinates under a name such as developer reference, end-user help, regulated disclosure or launch campaign. Rules use these coordinates to determine where they apply.
The same scope resolution supports guidance lookups, content checks and reuse of approved wording.
The axes a recipe declares
A project answers the five questions with named axes. Two come from the
recipe's structure: product is the profile a collection belongs to, and
channel is the surface it ships on. Both are derived from a collection's
channel: key.
You declare the rest. Two have names: brand (a workspace has brands, a brand
has products, a product ships channels) and mode, the kind of document at a
point in the Diátaxis sense: tutorial, how-to, reference, explanation. Any
other axis your content varies along goes under defaults.coordinates. A
language is an axis too.
Validity is set on a rule, not on a point: a rule can start and stop on a date. See the project model for the recipe keys.
Where to declare coordinates
Declare shared coordinates at project level. Collections inherit them and can override individual axes.
Coordinates are declared at the broadest true scope and inherited down; each level may override the one above it.
For example, a help-centre collection and an API-reference collection can inherit the same brand coordinate while declaring different channels. Files and passages cannot declare coordinates individually. To apply different guidance to a file, place it in a separate collection.
Rules inherit and expire
Rules apply within their declared scope and may include start and end dates. More specific rules can override inherited guidance.
For example, support replies may recognise a retired name during a transition period. A validity window limits that exception without requiring a later edit to remove it.
Build a first context from existing content
A context scan reads existing content, such as a repository, published site or style guide, and proposes axes, a voice profile and terms for review. You can edit the proposals before accepting them.
See Context scan for scan inputs and outputs, and Voice and corrections for maintaining guidance through reviewed corrections.
Approving governed changes
Governed changes require a change-set, a named draft of proposed edits. You can preview its effect on stored content and pilot it on selected content streams before merging it. Ordinary edits, such as definition changes, apply directly.
Corrections are evidence for new rules, and a reviewer decides which candidates to promote. See the Context hub for the approval process and Voice and corrections for candidate rules.
One graph, two scopes
kapi resolves context for local projects. Bowrain shares context across a workspace and adds collaborative review, access controls and cross-project analysis. Both use the same content engine. See kapi and Bowrain.
Where a language fits
Language is another coordinate, and it selects the terms and guidance that apply. The same model supports source-language checks and translation.