New Insight
No cards and no AI call for the empty Glyph. Now typing a short line, saving, then a longer one.
Cards section appeared (loading). Waiting for the request to complete.
Works: Explore next stays hidden until 12 words are saved, then generates from the text. Adding the test pin and docs, then committing.
Explore next now waits for 12 words, the same bar as descriptions. Pushed as 14d545c.
When the open Glyph (or a selection of only Glyphs) has fewer than 12 words, Explore next stays hidden and makes no AI call. Saving the body checks again, so the cards appear as soon as it reaches 12 words. A scope that includes a Node is not held back.
I checked it in the browser on a copy of the database:
A new empty Glyph made no Explore next request. “Just a short note.” (4 words), saved, still made none. After adding a sentence about cold brew (17 words) and saving, it made one request, and the cards were about that text.
The check runs in the browser, not the server. The /explore page asks the server for the same cards, and a server refusal would be remembered for that Glyph, so the cards would never come back after you write more. The catch is that another client calling the server directly isn’t held back.
Insights may have the same issue on an empty Glyph; I haven’t checked. Say if you want the same rule there.
Files changed: static/js/codexiq.js, tests/test_api.py, CHANGELOG.md What changed: as above. Documentation updated: changelog only; this follows on from the New Glyph fix. Changelog updated: yes, “Explore next waits until a Glyph has something to explore”. BACKLOG.md updated: no; it isn’t a backlog item. Verification performed: the browser checks above, and the teaser and dashboard tests (18 OK, including the new check). I didn’t rerun the full suite: this only adds code to one JS file. Diff summary: a word count over the Glyphs in scope; E
