fix(kiro): report real output tokens and stop discarding usable turns

open-sse/executors/kiro.js

@@ -144,6 +144,12 @@ function normalizeStopReason(value) {
   return reason || null;
 }
 
+// Of the reasons stopDisposition() folds into "terminal_incomplete", only these
+// mean "usable as far as it got, then the budget ran out" -- the case
+// finish_reason "length" exists for. cancelled / pause_turn are abandoned turns
+// whose partial content must stay private, so they are deliberately absent.
+const KIRO_TRUNCATION_STOP_REASONS = new Set(["model_context_window_exceeded", "max_tokens"]);
+
 function stopDisposition(stopReason, hasToolCalls) {
   if (["malformed_model_output", "invalid_model_output"].includes(stopReason)) return "retryable_protocol_failure";
   if (["cancelled", "pause_turn", "model_context_window_exceeded"].includes(stopReason)) return "terminal_incomplete";
@@ -711,7 +717,12 @@ export class KiroExecutor extends BaseExecutor {
     };
     const emitTools = (controller) => {
       for (const tool of state.tools.values()) {
-        const input = parsedToolInput(tool);
+        // Validate per tool, not per turn: one unusable fragment used to throw out
+        // of emitTools and take every other complete tool call in the same turn
+        // with it, which the client saw as a turn that answered nothing.
+        let input;
+        try {
+          input = parsedToolInput(tool);
           if (tool.name === "tool_call") {
             if (typeof input.name !== "string" || !input.name.trim()) {
               throw new Error("Invalid Kiro tool_call payload: missing nested MCP tool name");
@@ -720,6 +731,12 @@ export class KiroExecutor extends BaseExecutor {
               throw new Error("Invalid Kiro tool_call payload: missing nested MCP tool arguments");
             }
           }
+        } catch (error) {
+          state.droppedTools = (state.droppedTools || 0) + 1;
+          state.toolValidationError ||= error.message;
+          console.error(`[Kiro] dropping unusable tool call ${tool.id} (${tool.name}): ${error.message}`);
+          continue;
+        }
         const index = state.toolCounter++;
         emitDelta(controller, {
           tool_calls: [{
@@ -729,14 +746,26 @@ export class KiroExecutor extends BaseExecutor {
             function: { name: tool.name, arguments: "" }
           }]
         });
+        const serializedInput = JSON.stringify(input);
         emitDelta(controller, {
-          tool_calls: [{ index, function: { arguments: JSON.stringify(input) } }]
+          tool_calls: [{ index, function: { arguments: serializedInput } }]
         });
+        // Tool arguments are billed output like any other completion bytes. They
+        // were never added to totalContentLength, so the /4 estimator in finish()
+        // reported OUT 0 -- or the Math.max floor of 1 -- for every turn whose
+        // entire answer was a tool call.
+        state.totalContentLength += tool.name.length + serializedInput.length;
         state.hasToolCalls = true;
       }
       state.tools.clear();
       state.bufferedToolBytes = 0;
-      if (state.stopReason === "tool_use" && !state.hasToolCalls) {
+      // A declared tool turn that emitted no usable call is only fatal when the
+      // turn produced nothing else. Throwing unconditionally here escaped
+      // emitTools() with provenance "invalid_tool_call", which the integrity gate
+      // re-derived into a repair retry -- discarding text the client had already
+      // been promised.
+      if (state.stopReason === "tool_use" && !state.hasToolCalls &&
+          !state.hasText && !state.hasReasoning && !state.hasCode) {
         throw new Error("Kiro tool_use stop reason did not include a complete tool call");
       }
     };
@@ -796,7 +825,6 @@ export class KiroExecutor extends BaseExecutor {
         emitDelta(controller, { content: event.payload.content });
       } else if (eventType === "toolUseEvent") {
         state.sawToolUse = true;
-        if (state.toolValidationError) return true;
         const values = Array.isArray(event.payload) ? event.payload : [event.payload];
         if (!values[0]) throw new Error("Kiro toolUseEvent is empty");
         for (const value of values) {
@@ -924,9 +952,10 @@ export class KiroExecutor extends BaseExecutor {
         } catch (error) {
           const bufferExceeded = error.code === "KIRO_BUFFER_EXCEEDED";
           if (!bufferExceeded) {
+            // Keep whatever is already buffered: the rejected fragment belongs to
+            // one tool, and clearing the map dropped the complete calls too.
             state.toolValidationError ||= error.message;
-            state.tools.clear();
+            console.error(`[Kiro] tool fragment rejected, keeping ${state.tools.size} buffered tool(s): ${error.message}`);
-            state.bufferedToolBytes = 0;
             continue;
           }
           fail(
@@ -958,7 +987,16 @@ export class KiroExecutor extends BaseExecutor {
       }
       state.transportState = "clean_eof";
       const declaredDisposition = stopDisposition(state.stopReason, state.sawToolUse);
-      if (["retryable_protocol_failure", "terminal_incomplete", "terminal_refusal", "unknown_failure"].includes(declaredDisposition)) {
+      // model_context_window_exceeded / max_tokens map to terminal_incomplete. When
+      // they arrive after the model already streamed content, fail() threw away a
+      // complete-enough answer; a truncated turn is what finish_reason "length" is
+      // for. chunkIndex > 0 means at least one delta already reached the client.
+      const declaredTruncatedAfterOutput = declaredDisposition === "terminal_incomplete" &&
+        KIRO_TRUNCATION_STOP_REASONS.has(state.stopReason) && state.chunkIndex > 0;
+      if (declaredTruncatedAfterOutput) {
+        console.error(`[Kiro] truncated after ${state.chunkIndex} chunk(s) (stop_reason=${state.stopReason}); keeping output`);
+      }
+      if (!declaredTruncatedAfterOutput && ["retryable_protocol_failure", "terminal_incomplete", "terminal_refusal", "unknown_failure"].includes(declaredDisposition)) {
         const code = declaredDisposition === "retryable_protocol_failure"
           ? "kiro_retryable_protocol_failure"
           : declaredDisposition === "terminal_refusal"
@@ -975,16 +1013,6 @@ export class KiroExecutor extends BaseExecutor {
         );
         return;
       }
-      if (state.toolValidationError) {
-        fail(
-          controller,
-          "invalid_tool_call",
-          "invalid_kiro_tool_call",
-          state.toolValidationError,
-          { transport_state: state.transportState, stop_disposition: "retryable_protocol_failure" }
-        );
-        return;
-      }
       try {
         emitTools(controller);
       } catch (error) {
@@ -997,6 +1025,22 @@ export class KiroExecutor extends BaseExecutor {
         );
         return;
       }
+      // Fail only when the turn has nothing usable left. emitTools() validates
+      // per tool and drops just the unusable ones, so this has to run AFTER it:
+      // before, the rejected tool was still buffered and tools.size was never 0.
+      // A turn that also produced text keeps that text -- the dropped call is
+      // logged, not fatal.
+      if (state.toolValidationError && !state.hasToolCalls &&
+          !state.hasText && !state.hasReasoning && !state.hasCode) {
+        fail(
+          controller,
+          "invalid_tool_call",
+          "invalid_kiro_tool_call",
+          state.toolValidationError,
+          { transport_state: state.transportState, stop_disposition: "retryable_protocol_failure" }
+        );
+        return;
+      }
 
       const hasOutput = state.hasText || state.hasReasoning || state.hasCode || state.hasToolCalls;
       if (!hasOutput && !state.explicitStop) {
@@ -1011,7 +1055,13 @@ export class KiroExecutor extends BaseExecutor {
       }
 
       const disposition = stopDisposition(state.stopReason, state.hasToolCalls);
-      if (["retryable_protocol_failure", "terminal_incomplete", "terminal_refusal", "unknown_failure"].includes(disposition)) {
+      // Same reasoning as declaredTruncatedAfterOutput above.
+      const truncatedAfterOutput = disposition === "terminal_incomplete" &&
+        KIRO_TRUNCATION_STOP_REASONS.has(state.stopReason) && state.chunkIndex > 0;
+      if (truncatedAfterOutput) {
+        console.error(`[Kiro] truncated after ${state.chunkIndex} chunk(s) (stop_reason=${state.stopReason}); closing as length`);
+      }
+      if (!truncatedAfterOutput && ["retryable_protocol_failure", "terminal_incomplete", "terminal_refusal", "unknown_failure"].includes(disposition)) {
         const code = disposition === "retryable_protocol_failure"
           ? "kiro_retryable_protocol_failure"
           : disposition === "terminal_refusal"
@@ -1041,7 +1091,9 @@ export class KiroExecutor extends BaseExecutor {
           total_tokens: prompt + completion
         };
       }
-      const finishReason = state.hasToolCalls
+      const finishReason = truncatedAfterOutput
+        ? "length"
+        : state.hasToolCalls
           ? "tool_calls"
           : disposition === "length"
             ? "length"
@@ -1052,7 +1104,11 @@ export class KiroExecutor extends BaseExecutor {
       options.onTerminalState?.(diagnostics({
         terminal_provenance: state.terminalProvenance || "clean_eventstream_eof",
         transport_state: state.transportState,
-        stop_disposition: disposition
+        // Report what this exit actually did, not the raw disposition. The
+        // integrity gate re-derives its verdict from stop_disposition, so
+        // reporting "terminal_incomplete" for a turn we deliberately kept made
+        // it discard the very bytes we just released to the client.
+        stop_disposition: truncatedAfterOutput ? "length" : disposition
       }));
     };
 

open-sse/translator/request/claude-to-kiro.js

@@ -287,6 +287,18 @@ export function claudeToKiroRequest(model, body, stream, credentials) {
     toolSpecs,
     nameMap,
   });
+  // canonicalizeKiroConversation() already ran its second-chance repair (flatten
+  // every structured tool turn to text, then re-validate). A body that is STILL
+  // invalid here cannot be made shippable, and Kiro answers it with
+  // 400 {"message":"Improperly formed request.","reason":"REQUEST_BODY_INVALID"}.
+  // Fail locally instead: chatCore turns a falsy return into a 400 without
+  // spending an upstream call or a per-account cooldown. The taxonomy
+  // (role:N | pair:N | id:N | spec:N | orphan:0 | current) names the offending
+  // turn so the shape can be diagnosed from the log alone.
+  if (!canonical.valid) {
+    console.error(`[Kiro] refusing invalid conversation (claude → kiro): ${(canonical.errors || []).join(", ") || "unknown"} | turns=${(canonical.history || []).length + 1}`);
+    return null;
+  }
   const replayCurrent = canonical.currentMessage.userInputMessage;
   const userInputMessage = {
     content: replayCurrent.content || "",

open-sse/translator/request/openai-to-kiro.js

@@ -379,6 +379,18 @@ export function openaiToKiroRequest(model, body, stream, credentials) {
     toolSpecs,
     nameMap,
   });
+  // canonicalizeKiroConversation() already ran its second-chance repair (flatten
+  // every structured tool turn to text, then re-validate). A body that is STILL
+  // invalid here cannot be made shippable, and Kiro answers it with
+  // 400 {"message":"Improperly formed request.","reason":"REQUEST_BODY_INVALID"}.
+  // Fail locally instead: chatCore turns a falsy return into a 400 without
+  // spending an upstream call or a per-account cooldown. The taxonomy
+  // (role:N | pair:N | id:N | spec:N | orphan:0 | current) names the offending
+  // turn so the shape can be diagnosed from the log alone.
+  if (!canonical.valid) {
+    console.error(`[Kiro] refusing invalid conversation (openai → kiro): ${(canonical.errors || []).join(", ") || "unknown"} | turns=${(canonical.history || []).length + 1}`);
+    return null;
+  }
   const replayCurrent = canonical.currentMessage.userInputMessage;
 
   const payload = {

open-sse/translator/response/kiro-to-claude.js

@@ -75,6 +75,15 @@ export function kiroToClaudeResponse(chunk, state) {
         ? data.usage.completion_tokens
         : 0;
     state.usage = { input_tokens: promptTokens, output_tokens: outputTokens };
+    // Claude clients read cache_read/cache_creation to price a turn and to size
+    // their prompt cache. Both spellings are accepted because the Kiro executor
+    // emits the Chat shape and passthrough responses use the nested details form.
+    const cacheRead = data.usage.cache_read_input_tokens
+      ?? data.usage.prompt_tokens_details?.cached_tokens;
+    const cacheCreation = data.usage.cache_creation_input_tokens
+      ?? data.usage.prompt_tokens_details?.cache_creation_tokens;
+    if (typeof cacheRead === "number") state.usage.cache_read_input_tokens = cacheRead;
+    if (typeof cacheCreation === "number") state.usage.cache_creation_input_tokens = cacheCreation;
   }
 
   // First chunk → emit message_start.
@@ -254,6 +263,13 @@ export function kiroToClaudeNonStreaming(data) {
     usage: {
       input_tokens: usage.prompt_tokens || 0,
       output_tokens: usage.completion_tokens || 0,
+      // Same cache preservation as the streaming path above.
+      ...(typeof (usage.cache_read_input_tokens ?? usage.prompt_tokens_details?.cached_tokens) === "number"
+        ? { cache_read_input_tokens: usage.cache_read_input_tokens ?? usage.prompt_tokens_details.cached_tokens }
+        : {}),
+      ...(typeof (usage.cache_creation_input_tokens ?? usage.prompt_tokens_details?.cache_creation_tokens) === "number"
+        ? { cache_creation_input_tokens: usage.cache_creation_input_tokens ?? usage.prompt_tokens_details.cache_creation_tokens }
+        : {}),
     },
   };
 }

tests/unit/kiro-usage-and-tool-integrity.test.js

@@ -0,0 +1,369 @@
+/**
+ * Five Kiro defects, all on the OAuth/social Kiro route (kr/claude-sonnet-4.5
+ * with a >100k context). Production shape: 402 of 2156 usageHistory rows for
+ * Kiro recorded completionTokens 0, and the 25 newest rows all sat pinned at
+ * exactly 1 output token against prompts of 80k-103k -- i.e. the router was not
+ * measuring the answer, it was measuring nothing and rounding up.
+ *
+ * A. OUT 0 / OUT 1. finish() estimates completion tokens as
+ *    totalContentLength / 4, but tool-call bytes were never added to
+ *    totalContentLength. A turn whose entire answer is a tool call therefore
+ *    measured as an empty answer (Math.max(1, ...) is where the 1 comes from).
+ *
+ * B. Truncation threw away a complete-enough answer. stopDisposition() maps
+ *    model_context_window_exceeded (and max_tokens alongside tool calls) to
+ *    terminal_incomplete, which hard-fails the turn -- even when the model had
+ *    already streamed text. A truncated turn is what finish_reason "length" is
+ *    for. Both disposition gates needed the bypass: the declared-stop-reason
+ *    gate runs first and returns, so patching only finish() would be dead code.
+ *
+ * C. One bad tool fragment killed every good one. Three separate latches:
+ *    emitTools() validated per turn and threw out of the loop; the frame-loop
+ *    catch cleared state.tools wholesale; and the toolUseEvent branch returned
+ *    early forever once toolValidationError was set. Net effect for the client:
+ *    a turn that answered nothing.
+ *
+ * D. Cache tokens dropped on the kiro:claude route. kiro-to-claude built usage
+ *    from prompt_tokens/completion_tokens only, so Claude clients lost
+ *    cache_read_input_tokens / cache_creation_input_tokens and could neither
+ *    price the turn nor size their prompt cache.
+ *
+ * E. Defence-in-depth only: both request translators discarded canonical.valid.
+ *    canonicalizeKiroConversation() self-heals every failure mode it can detect
+ *    (see the test below), so the guard is unreachable by construction today --
+ *    it exists so a future validator rule cannot ship an unusable body silently.
+ *
+ * Kiro also answers an unusable conversation with 400 {"message":"Improperly
+ * formed request.","reason":"REQUEST_BODY_INVALID"}, which cools down every
+ * account that reports it. That is a property of ERROR_RULES rather than of this
+ * executor -- the same body fails identically on any account -- so it belongs to
+ * the request-scoped `fallback: false` rule kind, not here.
+ */
+import { afterEach, beforeEach, describe, expect, it, vi } from "vitest";
+
+const fetchMock = vi.fn();
+
+vi.mock("../../open-sse/utils/proxyFetch.js", () => ({
+  proxyAwareFetch: (...args) => fetchMock(...args)
+}));
+
+const { KiroExecutor } = await import("../../open-sse/executors/kiro.js");
+const { kiroToClaudeResponse, kiroToClaudeNonStreaming } = await import(
+  "../../open-sse/translator/response/kiro-to-claude.js"
+);
+const { validateKiroConversation, canonicalizeKiroConversation } = await import(
+  "../../open-sse/translator/concerns/kiroConversation.js"
+);
+
+const encoder = new TextEncoder();
+const credentials = {
+  accessToken: "test-token",
+  providerSpecificData: { kiroToolCallRepair: true }
+};
+
+function crc32(bytes) {
+  let crc = 0xffffffff;
+  for (const byte of bytes) {
+    crc ^= byte;
+    for (let bit = 0; bit < 8; bit++) {
+      crc = (crc >>> 1) ^ ((crc & 1) ? 0xedb88320 : 0);
+    }
+  }
+  return (crc ^ 0xffffffff) >>> 0;
+}
+
+function encodeHeader(name, value) {
+  const nameBytes = encoder.encode(name);
+  const valueBytes = encoder.encode(value);
+  const bytes = new Uint8Array(1 + nameBytes.length + 3 + valueBytes.length);
+  let offset = 0;
+  bytes[offset++] = nameBytes.length;
+  bytes.set(nameBytes, offset);
+  offset += nameBytes.length;
+  bytes[offset++] = 7;
+  new DataView(bytes.buffer).setUint16(offset, valueBytes.length, false);
+  offset += 2;
+  bytes.set(valueBytes, offset);
+  return bytes;
+}
+
+function concat(chunks) {
+  const output = new Uint8Array(chunks.reduce((size, chunk) => size + chunk.byteLength, 0));
+  let offset = 0;
+  for (const chunk of chunks) {
+    output.set(chunk, offset);
+    offset += chunk.byteLength;
+  }
+  return output;
+}
+
+function checksum(bytes) {
+  const view = new DataView(bytes.buffer, bytes.byteOffset, bytes.byteLength);
+  view.setUint32(8, crc32(bytes.subarray(0, 8)), false);
+  view.setUint32(bytes.byteLength - 4, crc32(bytes.subarray(0, bytes.byteLength - 4)), false);
+  return bytes;
+}
+
+function frameFromEntries(entries, payload) {
+  const headers = concat(entries.map(([name, value]) => encodeHeader(name, value)));
+  const payloadBytes = encoder.encode(JSON.stringify(payload));
+  const totalLength = 12 + headers.byteLength + payloadBytes.byteLength + 4;
+  const frame = new Uint8Array(totalLength);
+  const view = new DataView(frame.buffer);
+  view.setUint32(0, totalLength, false);
+  view.setUint32(4, headers.byteLength, false);
+  frame.set(headers, 12);
+  frame.set(payloadBytes, 12 + headers.byteLength);
+  return checksum(frame);
+}
+
+function frame(eventType, payload) {
+  return frameFromEntries([[":event-type", eventType]], payload);
+}
+
+function response(frames, status = 200) {
+  return new Response(new ReadableStream({
+    start(controller) {
+      for (const value of frames) controller.enqueue(value);
+      controller.close();
+    }
+  }), { status, statusText: status === 200 ? "OK" : "Upstream Error" });
+}
+
+async function execute(executor = new KiroExecutor(), overrides = {}) {
+  return executor.execute({
+    model: "kr/claude-opus-4.8",
+    body: { systemPrompt: "base", conversationState: {} },
+    stream: true,
+    credentials,
+    ...overrides
+  });
+}
+
+// Output is held behind the ": kiro-validation" heartbeat until clean EOF, so
+// every executor assertion has to drain the whole response.
+async function run(frames) {
+  fetchMock.mockResolvedValueOnce(response(frames));
+  return await (await execute()).response.text();
+}
+
+// Same as run(), but with the bounded tool-call repair retry disabled, so a
+// hard failure is surfaced from the first attempt instead of triggering a
+// second upstream fetch.
+async function runNoRepair(frames) {
+  fetchMock.mockResolvedValueOnce(response(frames));
+  const result = await execute(new KiroExecutor(), {
+    credentials: { accessToken: "test-token", providerSpecificData: { kiroToolCallRepair: false } }
+  });
+  return await result.response.text();
+}
+
+// The estimator only runs when metering and context usage both arrived and the
+// upstream reported no token totals of its own -- the exact production shape.
+const METERED = [
+  frame("meteringEvent", { usage: 2, unit: "credit" }),
+  frame("contextUsageEvent", { contextUsagePercentage: 10 })
+];
+
+function usageFrom(body) {
+  const usages = body
+    .split("\n")
+    .filter(line => line.startsWith("data: ") && line.includes('"usage"'))
+    .map(line => JSON.parse(line.slice(6)).usage)
+    .filter(Boolean);
+  return usages[usages.length - 1];
+}
+
+beforeEach(() => {
+  fetchMock.mockReset();
+  vi.spyOn(console, "error").mockImplementation(() => {});
+});
+afterEach(() => {
+  vi.restoreAllMocks();
+});
+
+describe("A: tool-call bytes count as output tokens", () => {
+  it("does not report a tool-only turn as OUT 1", async () => {
+    const body = await run([
+      frame("toolUseEvent", {
+        toolUseId: "call_a",
+        name: "tool_call",
+        input: { name: "mcp_search", arguments: { query: "why is the router reporting zero output" } }
+      }),
+      frame("metadataEvent", { stopReason: "tool_use" }),
+      ...METERED
+    ]);
+
+    const usage = usageFrom(body);
+    expect(usage).toBeDefined();
+    // Was 1: the Math.max floor over a totalContentLength of 0.
+    expect(usage.completion_tokens).toBeGreaterThan(10);
+    expect(usage.total_tokens).toBe(usage.prompt_tokens + usage.completion_tokens);
+  });
+
+  it("still counts plain text output", async () => {
+    const body = await run([
+      frame("assistantResponseEvent", { content: "x".repeat(400) }),
+      frame("metadataEvent", { stopReason: "end_turn" }),
+      ...METERED
+    ]);
+    expect(usageFrom(body).completion_tokens).toBe(100);
+  });
+});
+
+describe("C: one unusable tool fragment does not take the whole turn with it", () => {
+  it("ships the valid call and drops only the invalid one", async () => {
+    const body = await run([
+      frame("toolUseEvent", {
+        toolUseId: "good",
+        name: "tool_call",
+        input: { name: "mcp_search", arguments: { q: "router" } }
+      }),
+      // No nested MCP name -> unusable, cannot be forwarded to the client.
+      frame("toolUseEvent", { toolUseId: "bad", name: "tool_call", input: { arguments: { q: "router" } } }),
+      frame("metadataEvent", { stopReason: "tool_use" }),
+      ...METERED
+    ]);
+
+    expect(body).toContain('\\"name\\":\\"mcp_search\\"');
+    expect(body).not.toContain('"id":"bad"');
+    expect(body).toContain('"finish_reason":"tool_calls"');
+  });
+
+  it("keeps streamed text when the only tool call is unusable", async () => {
+    const body = await run([
+      frame("assistantResponseEvent", { content: "Here is what I found." }),
+      frame("toolUseEvent", { toolUseId: "bad", name: "tool_call", input: { arguments: {} } }),
+      frame("metadataEvent", { stopReason: "tool_use" }),
+      ...METERED
+    ]);
+
+    expect(body).toContain("Here is what I found.");
+    expect(body).not.toContain("invalid_kiro_tool_call");
+  });
+
+  it("still hard-fails when nothing usable was produced at all", async () => {
+    const body = await runNoRepair([
+      frame("toolUseEvent", { toolUseId: "bad", name: "tool_call", input: { arguments: {} } }),
+      frame("metadataEvent", { stopReason: "tool_use" })
+    ]);
+    expect(body).toContain("invalid_kiro_tool_call");
+    expect(fetchMock).toHaveBeenCalledTimes(1);
+  });
+});
+
+describe("B: truncation after output closes as length, not as a failure", () => {
+  it("keeps the text and reports finish_reason length", async () => {
+    const body = await run([
+      frame("assistantResponseEvent", { content: "Partial but usable answer." }),
+      frame("metadataEvent", { stopReason: "model_context_window_exceeded" }),
+      ...METERED
+    ]);
+
+    expect(body).toContain("Partial but usable answer.");
+    expect(body).toContain('"finish_reason":"length"');
+    expect(body).not.toContain("kiro_terminal_incomplete");
+  });
+
+  it("still fails a truncation that produced nothing", async () => {
+    const body = await run([
+      frame("metadataEvent", { stopReason: "model_context_window_exceeded" })
+    ]);
+    expect(body).toContain("kiro_terminal_incomplete");
+  });
+});
+
+describe("D: cache tokens survive the kiro -> claude translation", () => {
+  const finishChunk = (usage) => ({
+    choices: [{ index: 0, delta: {}, finish_reason: "stop" }],
+    usage
+  });
+
+  function finalUsage(usage) {
+    const state = {};
+    // Usage rides an earlier chunk in the real stream; feed it the same way.
+    kiroToClaudeResponse({ choices: [{ index: 0, delta: { content: "hi" } }], usage }, state);
+    const events = kiroToClaudeResponse(finishChunk(usage), state) || [];
+    return events.find(e => e.type === "message_delta")?.usage;
+  }
+
+  it("forwards the flat Chat spelling the executor emits", () => {
+    expect(finalUsage({
+      prompt_tokens: 103000,
+      completion_tokens: 640,
+      cache_read_input_tokens: 98000,
+      cache_creation_input_tokens: 1912
+    })).toEqual({
+      input_tokens: 103000,
+      output_tokens: 640,
+      cache_read_input_tokens: 98000,
+      cache_creation_input_tokens: 1912
+    });
+  });
+
+  it("also accepts the nested details spelling used on passthrough", () => {
+    expect(finalUsage({
+      prompt_tokens: 500,
+      completion_tokens: 20,
+      prompt_tokens_details: { cached_tokens: 480, cache_creation_tokens: 20 }
+    })).toEqual({
+      input_tokens: 500,
+      output_tokens: 20,
+      cache_read_input_tokens: 480,
+      cache_creation_input_tokens: 20
+    });
+  });
+
+  it("omits the cache keys when the upstream reported none", () => {
+    expect(finalUsage({ prompt_tokens: 500, completion_tokens: 20 }))
+      .toEqual({ input_tokens: 500, output_tokens: 20 });
+  });
+
+  it("preserves cache on the non-streaming path too", () => {
+    const message = kiroToClaudeNonStreaming({
+      choices: [{ message: { content: "hi" } }],
+      usage: { prompt_tokens: 90, completion_tokens: 4, cache_read_input_tokens: 80 }
+    });
+    expect(message.usage).toMatchObject({
+      input_tokens: 90,
+      output_tokens: 4,
+      cache_read_input_tokens: 80
+    });
+    expect(message.usage).not.toHaveProperty("cache_creation_input_tokens");
+  });
+});
+
+describe("E: the translator valid-guard is defence-in-depth", () => {
+  const SPECS = [{ toolSpecification: { name: "read_file", inputSchema: { json: { type: "object" } } } }];
+
+  it("validateKiroConversation names the offending turn", () => {
+    // Hand-built, NOT normalized: assistant first, then a tool call with no
+    // matching result and a name no spec declares.
+    const result = validateKiroConversation(
+      [{ assistantResponseMessage: { content: "hi", toolUses: [{ toolUseId: "t1", name: "ghost" }] } }],
+      { userInputMessage: { content: "go" } },
+      SPECS
+    );
+    expect(result.valid).toBe(false);
+    expect(result.errors).toContain("role:0");
+    expect(result.errors).toContain("pair:0");
+    expect(result.errors).toContain("spec:0");
+  });
+
+  it("canonicalizeKiroConversation heals that same conversation", () => {
+    // This is why the guard cannot fire today: normalizeTurns() forces
+    // user-first/user-last alternation and non-empty content, and the
+    // second-chance pass flattens every structured tool turn to text.
+    const out = canonicalizeKiroConversation({
+      history: [{ assistantResponseMessage: { content: "hi", toolUses: [{ toolUseId: "t1", name: "ghost" }] } }],
+      currentMessage: { userInputMessage: { content: "" } },
+      modelId: "claude-sonnet-4-5",
+      toolSpecs: SPECS
+    });
+
+    expect(out.valid).toBe(true);
+    expect(out.errors).toEqual([]);
+    expect(out.currentMessage.userInputMessage.content).toBe("continue");
+    expect(out.history[0].userInputMessage).toBeDefined();
+  });
+});