lsm / neokai / 28486988990

82%

push

github

fix(openai-bridge): cap 4xx log noise, use histograms, skip body summary for quota/auth errors (#2217)

summarizeResponsesRequestFor4xx now reports input item types and per-message
content block types as histograms (counts per type) instead of full arrays, so
a long session with hundreds of tool calls can't balloon the 4xx log line.

logUpstream4xx no longer dumps the translated request summary for 429/401/403
(rate-limit/quota/auth) — those aren't body problems — and emits a short line
instead. 400/422/other 4xx still log the full summary, with the JSON sliced to
1000 chars to bound line size. Error propagation is unchanged.

11446 of 16495 branches covered (69.39%)

Branch coverage included in aggregate %.

17 of 17 new or added lines in 1 file covered. (100.0%)

97766 of 115954 relevant lines covered (84.31%)

301.0 hits per line