| text | expected | status |
|---|---|---|
| Guest | 'guest' | ✓ |
| Order received | 'order-received' | ✓ |
| Work Object | 'work-object' | ✓ |
| do_the_thing | 'do-the-thing' | ✓ |
| Buy / sell | 'buy-sell' | ✓ |
| Guest #1 | 'guest-1' | ✓ |
| café | 'caf' | ✓ |
| booking system | 'booking-system' | ✓ |
| text | status |
|---|---|
| --- | ✓ |
| ✓ | |
| ✓ | |
| ### | ✓ |
| phase_name | status |
|---|---|
| given | ✓ |
| then | ✓ |
| label_kind | status |
|---|---|
| deferred-template | ✓ |
| t-string | ✓ |
| not-a-string | ✓ |
| phase_name | status |
|---|---|
| given | ✓ |
| when | ✓ |
| then | ✓ |
| text | status |
|---|---|
| count={obj.attr} | ✓ |
| {d[key]} | ✓ |
| {x + 1} | ✓ |
| phase | step body | status |
|---|---|---|
| given | ✓ | |
| when | ✓ | |
| preset | url_scheme | status |
|---|---|---|
| vscode | vscode://file/{path}:{line} | ✓ |
| cursor | cursor://file/{path}:{line} | ✓ |
| zed | zed://file/{path}:{line} | ✓ |
| pycharm | pycharm://open?file={path}&line={line} | ✓ |
ContextVar so a nested run can displace and restore it.--given-lint pass that checks recorded scenarios against the Lint rule catalog once the Report model is built. It enforces the structural subset of the narration conventions — it cannot check whether a Step's text is semantically true.@scenario(...). Not every pytest test is a scenario — undecorated tests are tolerated but not collected.with given(...) / when(...) / then(...) block, or the root recording from a step fixture. Steps nest, and each carries a phase, narration, attachments and children — but no status or error of its own: a failure is recorded on the Scenario, and per Case on the Parameter table, which is where both renderers read it.given, when, or then. A step has exactly one phase.@scenario(name, tags=[...]). Used by the report's filter UI, where a / nests it (ticket/ABC-123 under ticket) and selecting a prefix filters to every tag beneath it.attach(label, content).@scenario-decorated test that also carries @pytest.mark.parametrize(...). Produces multiple scenario records during a run; pytest-given groups them.param (a @pytest.mark.parametrize input), a derived (a narrated value that varies across cases) or an attachment (an attachment whose payload varies). Appears in the report below the grouped-template steps.@given(text). Only @given is allowed on fixtures; @when / @then are rejected.with given(...), popped on exit. Mirrored inside a fixture recording while a fixture body is running.runtime rule reads the recorded scenarios; an ast rule parses the step bodies' source. The catalog in lint/base.py is data, so severities, config validation and docs stay in sync with one table.Glossary(), with .actor(...), .work_object(...), .verb(...) registration methods and g('foo') (declare-or-get a kindless term, optional definition=) / g['foo'] (get-only, raises on unknown) accessor forms.story('Title', [activity(...), ...]). Stories are first-class report tabs and the unit of coverage.actor + verb + work_object plus optional connective words. Constructed by activity(...).@scenario(group_parametrized=False) declines the merge, so each case lives on as its own scenario.Narration bundles the flat rendered text with parts — empty for plain-string authoring, and for a t-string or pytest_given.Template a list of NarrationLiteral / NarrationValue / NarrationPlaceholder / NarrationTermRef pieces. The structured form lets the templatizer and renderer treat parametrize-bound values specially without regex tricks.when action and its then outcome as two sibling steps from a single with block. Used mainly to narrate an expected raise (with when_then('the action', 'the error is raised'), pytest.raises(...)), so the action and its outcome stay distinct steps.{name} placeholder or attachment badge pointing at a parameter-table column. The baseline tree comes from the first passed case, and every other passed case must narrate that same template.tests/test_x.py::test_y[a-b]). Used as a key throughout the collector.price * 1.2). One that varies becomes a derived column instead.given_source_link config — a preset name like vscode / github, or a raw URL template. Captured as a SourceLocation (POSIX relpath + 1-indexed line) from pytest.Item.location for a scenario, from the declaration site for a Story or Term. Disabled by default.light, dark, or auto (follow the viewer's system). The given_theme config sets the default a report opens in; a viewer's own choice from the report's theme control, once made, wins over it in that browser.off, warn or error — defaulted by the catalog and overridable per rule via given_lint_rules. Only error fails the run; warn reports in the terminal summary; an off rule is not evaluated at all.FileGlossary(path) class. It parses all GFM pipe tables in the file into the same inner Glossary model; terms are accessed by name (g['Guest'], case-insensitive). Kind inference fills in term kinds post-collection from activity slot positions when no explicit kind_column is configured.g('foo') and g['foo'] on a code glossary return, and every file glossary lookup. The handle is the same TermHandle the typed registrations (g.actor(...) and friends) hand back; declared_kind is None is what marks the deferral. The deferral is in the handing over, not the term: a row with an explicit kind_column arrives through the same handle already kinded, while the rest stay None until kind inference runs.None when kindless), and an optional definition (`strTermHandle, the same type from every accessor (g.actor(...), g('foo'), g['Guest'], a captured guest = ...). It is what steps, scenario titles and activities interpolate, in one of three surface forms: bare (room — the canonical display), `.low` (room.low — lowercased), or called (room('Deluxe Suite') — an instance on an Actor or Work Object, an inflection on a Verb). Each form renders as a term ref.book('books') records books as a surface form of the canonical book.NarrationTermRef in step text and as ActivityTermRef inside activity prose.guest('Alice') is an instance of the Guest actor). Instances aggregate in the Glossary tab's refs block.ActivityPart = ActivityTermRef or ActivityWord): a reference to a glossary term, whose kind resolves via the glossary, or a bare path word — a node label or edge connective that carries no kind or id, is never classified by inference, and never reaches the glossary. That last is what separates it from a kindless term, which has an id and is tracked.path(...) lets alternate activity sequences share a prefix.@scenario(story=, activities=) and the activity= kwarg on given/when/then.activity= pin, regardless of its narration or its term count.infer_glossary_kinds) that assigns each undeclared term a kind from the slot positions it occupies across all story activities; declared kinds are verified against observed positions instead. A term used in incompatible slots (or conflicting with its declared kind) raises. A term never referenced by any activity stays kindless.kind=None — left unset when kind inference finds no story slot to infer from (a term used only in t-string steps, never in an activity). Shown in the report's Uncategorized bucket.definition is None; surfaced by a badge and filter in the Glossary view. Orthogonal to kindless.