Dashboard › opencode-lore › Distillation
7cc559dc-e0ed-48a5-99a9-8ea61447ca45["lore_tm_v1_fXHDF91bhLkZEjA9-JBnqVicDmOSsjnROOnpvA-SLmA"]
Date: Sep 16, 2026
packages/gateway/test/openai-responses-recall-aware-stream.test.ts added responseEvents(output), which parses SSE blocks matching /^event: (.+)\ndata: (.+)$/gm, JSON-decodes each data payload, retains only non-array objects, and silently skips malformed JSON or non-object payloads.packages/gateway/test/openai-responses-recall-aware-stream.test.ts added expectVisibleOutputIndicesToMatchTerminal(output): it finds the response.completed event, requires response.output to be defined, then checks every visible event with a safe-integer output_index and an item ID from data.item_id or data.item.id against terminalResponse.output[output_index].id."fc_sparse_terminal_recall" at output_index: 9; the test drains the client and checks emitted response.completed events."places continuation after deferred principal output without index collisions" uses principal recallCall(0, { query: "architecture" }) and continuation textItem(0, "continued answer", "msg_sparse_followup"); it verifies one recall and collision-free placement in the projected output."already visible" and item ID "msg_dense_visible"; assertions inspect parsed SSE events and reject an unwanted event shape with expect(events).not.toContainEqual(...)."allows deferred sparse output before a higher-index recall" with deferred function-call output at output_index: 0 and recallCall(9, { query: "private query" }, "fc_late_deferred_recall"); assertions inspect parsed events and verify prohibited recall-related output is absent.response.custom, follow-up response ID "resp_allowlisted_followup", principal response ID "resp_allowlisted_principal", model "gpt-5.6-terra", and onRecall: async ({ query }) => (...); cases distinguish eventSecret and terminalSecret handling.