andgineer / echo-words

96%

push

github

Version v1.14.0
 README and docs: mention looking a word up from your own language
 Display specs: the blind counts stand as graded, the paid step on no answer
 Display: the bold reading is pinned on article lines models really wrote
 Review loop: a test gap is a must-fix only when it changes a recorded answer
 Display specs: Markdown bold reads as bold, a typographic-quoted lookup is read
 Display: each depth check of a Markdown bold, and its symbol class, is pinned by a test
 Cost requirement: every situation that reaches the paid model is named
 Display: a Markdown bold's tag and punctuation guards are each pinned by a test, and a long answer is read in linear time
 Display: a Markdown bold neither crosses a tag nor pairs past punctuation, and a bracket beside it stays
 Display: Markdown bold reads as bold, a stray bracket around it goes, a curly-quoted lookup is read
 Configuration docs: a lookup the app tries on its own is dropped, not bought
 Reverse lookup: the app's own setting is tested, and the log says which answer stood
 Reverse lookup: a lookup the pool's main model did not answer is answered by the paid model
 Every change goes through the orchestration loop
 Triage: one plan replaces the defect list; the rest accepted

1 of 1 new or added line in 1 file covered. (100.0%)

3963 of 4119 relevant lines covered (96.21%)

0.96 hits per line