Overview

7 days30 days

Latest CI Pipeline Executions

Fuzzy

Succeeded
readme-header-endpoint
a2b0f1bd Merge 72359f934c3328a8a85fcd3dbf8b898d074ca31b into f7f639c3947ea14842ae89db10c0e824fc46f220
by Alem Tuzlak
Succeeded
readme-header-endpoint
a2b0f1bd fix(scripts): sync-models array insertion breaks on single-line arrays (#1042) addToArray inserted new entries just before the closing bracket, assuming the last existing entry ended with a trailing comma — true for every provider's long, oxfmt-multi-lined model array, but not for grok's two-entry GROK_CHAT_MODELS, which sat on one line. The day OpenRouter listed grok-4.5 the generator emitted export const GROK_CHAT_MODELS = [GROK_BUILD_0_1.name, GROK_4_3.name GROK_4_5.name, ] as const — a syntax error that made the pipeline's closing `pnpm format` exit 2. The daily sync-models workflow has failed on it every run since July 9, stranding 11 new models across four providers (grok-4.5, two Geminis, the GPT-5.6 trio, claude-opus-5/-fast). Rewrite the insertion to be correct by construction: insert immediately AFTER the opening bracket, so the script only ever writes separators it owns (each inserted line carries its own trailing comma) and the existing body — trailing commas or not, section comments, or empty — follows unchanged, with no guessing about its shape. New models therefore land at the top of each array; `pnpm format` reflows the result. The function also loses its dotAll regex for a plain indexOf. Verified by running the full `pnpm generate:models` pipeline: all 11 models insert, oxfmt passes (the previously failing step), OpenAI's in-array section comments survive, and ai-grok / ai-gemini / ai-openai typecheck with the generated entries. Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
by Tom Beckenham