ai-doc-parity — fixtures transcribed from a document, NOT recorded
The files of this directory are not answers recorded from a server. They are the example bodies printed in the
AI workstream contract (API-NEXT-AI.md, §5 Smart Inbox and §8 Research), copied verbatim, because the scratch
backend used to record the other groups (the foundation build) does not have these routes: it answers the HTML 404
page of Next.js (see ../inbox/suggestions-cached-only.txt, ../research/export-tex.txt) or ignores the parameter
(../research/ref-list-ids.json).
| file | source in API-NEXT-AI.md |
|---|---|
inbox-suggestions.json |
§5.1, body of GET /api/inbox/<id>/suggestions |
inbox-apply.json |
§5.2, body of POST /api/inbox/<id>/apply |
inbox-apply-nothing.json |
§5.2, the 400 nothing_to_apply body |
references-ids.json |
§8.3, body of GET /api/research/references?ids= (with its x-missing-ids header; the … are in the document) |
export-tex-report.json |
§8.1, value of the x-export-report header |
export-tex-bad-id.json |
§8.1, the 400 bad_id body |
_index.json has the format written by Scripts/spn_recorder.py, so NextFixture loads them the same way.
Re-record them from a server that has the AI workstream as soon as one is available, and delete this directory.