Dashboard › opencode-lore › Distillation
181eb166-51c2-460c-bd52-7a0e78bed246["lore_tm_v1_KrL442OCBTZc2SObvaw93nmrj5Be6TQDD_y_B532qxM","lore_tm_v1_Mi2Wmt24wAtDLI_wZwbcEP3hyKvMsiqulwACBpWExpE","lore_tm_v1_ZsC3MvUMBAcKM9Dnppx8XZmjJLBvisZTq-XZkfqv9HM","lore_tm_v1_3_eygD6nfc7PenHXmfhMKNkELsYaGvIKPvUeePPFv_0"]
Date: Sep 16, 2026
--printConsoleTrace.vitest/4.1.8; its CLI exposes --printConsoleTrace for always printing console stack traces./home/byk/Code/opencode-lore-responses-reference-v2 as PID 482674, scanning /proc for processes whose working directory was under that repository and whose command matched node, vitest, tsx, or pnpm; it produced no matching output.1194ms.packages/gateway/test/openai-responses-recall-aware-stream.test.ts exercised recall-aware streaming failure paths: malformed continuation after unfinished recall arguments logged category=follow_up_protocol; client disconnect after continuation delivery began logged category=delivery; late reader from abort-ignoring follow-up setup, cancellation racing a dropped continuation, and cancellation with non-settling follow-up setup logged category=follow_up_setup; failed recall-only continuation logged category=follow_up_failed.runPrincipalAttempt at packages/gateway/src/pipeline.ts:11512 and packages/gateway/src/pipeline.ts:11635; outer OpenAI Responses recall-aware stream errors were logged at packages/gateway/src/pipeline.ts:11761; logging passed through packages/core/src/log.ts:499.packages/core/test/embedding.test.ts completed with 61 tests, including 6 skipped, in 1112ms.packages/core/test/embedding-init-retry.test.ts verified that local embedding initialization latches permanently only after the retry budget is exhausted. The simulated third failure reported ERROR_CODE: 7, Failed to load model because protobuf parsing failed, through LocalProvider.handleInitError at packages/core/src/embedding/local.ts:1008/:1005, the worker listener at packages/core/src/embedding/local.ts:739, and failInit at packages/core/test/embedding-init-retry.test.ts:108; the test invocation was at line 254.packages/core/test/semantic-lint.test.ts failure was checkInvariants (funnel, stubbed LLM) > fans a violation out to all near-duplicate hunks with ONE judge call: at line 1233, prompt was expected to be called exactly once but was called 0 times; the intended behavior was one judge call for the cluster representative with the verdict fanned out to both files.