nolo-cli 0.1.41 → 0.1.43

package/client/agentRun.test.ts

@@ -2,7 +2,7 @@ import { describe, expect, test } from "bun:test";
 import { Writable } from "node:stream";
 
 import {
-  classifyTaskRunReviewStatus,
+  classifyTaskContextReviewStatus,
   findServerPlatformTools,
   runAgentTurn,
 } from "./agentRun";
@@ -24,14 +24,14 @@ class CaptureOutput extends Writable {
 
 describe("cli agent run client", () => {
   test("classifies only clear reviewer outcomes", () => {
-    expect(classifyTaskRunReviewStatus("Review decision: passed")).toBe("passed");
-    expect(classifyTaskRunReviewStatus("Review decision: needs_changes")).toBe("needs_changes");
-    expect(classifyTaskRunReviewStatus("Review decision: blocked")).toBe("blocked");
-    expect(classifyTaskRunReviewStatus("LGTM, no issues found")).toBe("passed");
-    expect(classifyTaskRunReviewStatus("Changes requested for missing test coverage")).toBe(
+    expect(classifyTaskContextReviewStatus("Review decision: passed")).toBe("passed");
+    expect(classifyTaskContextReviewStatus("Review decision: needs_changes")).toBe("needs_changes");
+    expect(classifyTaskContextReviewStatus("Review decision: blocked")).toBe("blocked");
+    expect(classifyTaskContextReviewStatus("LGTM, no issues found")).toBe("passed");
+    expect(classifyTaskContextReviewStatus("Changes requested for missing test coverage")).toBe(
       "needs_changes"
     );
-    expect(classifyTaskRunReviewStatus("I looked at the diff and have notes")).toBeUndefined();
+    expect(classifyTaskContextReviewStatus("I looked at the diff and have notes")).toBeUndefined();
   });
 
   test("identifies server platform tools that local runtime cannot provide", () => {
@@ -111,18 +111,23 @@ describe("cli agent run client", () => {
           reason: "connector_execution_failed",
           code: "connector_disconnected_mid_run",
           message: "Connector disconnected for machine: machine-win",
+          dialogId: "dialog-failed-1",
         }, { status: 502 }),
     });
 
-    expect(result).toEqual({ exitCode: 1 });
+    expect(result).toEqual({ exitCode: 1, dialogId: "dialog-failed-1" });
     expect(output.text()).toContain("[nolo] Agent request failed: HTTP 502");
     expect(output.text()).toContain("Connector execution failed");
     expect(output.text()).toContain("Connector disconnected for machine: machine-win");
     expect(output.text()).toContain("code=connector_disconnected_mid_run");
     expect(output.text()).toContain("reason=connector_execution_failed");
+    expect(output.text()).toContain("[nolo] failed dialog: dialog-failed-1");
+    expect(output.text()).toContain(
+      'nolo agent run agent-win-codex --continue dialog-failed-1 --msg "retry"'
+    );
   });
 
-  test("prepends task-run context for HTTP agent runs", async () => {
+  test("converts task row context to subjectRefs without legacy prompt or runtime taskRun payload", async () => {
     const output = new CaptureOutput();
     const requests: Array<{ body: any }> = [];
 
@@ -135,10 +140,8 @@ describe("cli agent run client", () => {
       env: { AUTH_TOKEN: "token-123" },
       output,
       runtimeMode: "server",
-      taskRunContext: {
+      taskRowContext: {
         rowDbKey: "row-b2e06f801f-01TASK",
-        taskRunId: "taskrun-1",
-        workItemId: "frontend-filter",
         artifactIds: ["artifact-1"],
       },
       fetchImpl: async (_url, init) => {
@@ -147,22 +150,17 @@ describe("cli agent run client", () => {
       },
     });
 
-    expect(requests[0]?.body.userInput).toContain("AI-native task-run context:");
-    expect(requests[0]?.body.userInput).toContain("current task board is only the present observation and validation use case");
-    expect(requests[0]?.body.userInput).toContain("rowDbKey: row-b2e06f801f-01TASK");
-    expect(requests[0]?.body.userInput).toContain("workItemId: frontend-filter");
-    expect(requests[0]?.body.userInput).toContain("User task:\nFix the filter UI");
-    expect(requests[0]?.body.runtimeContext.taskRun).toEqual({
-      rowDbKey: "row-b2e06f801f-01TASK",
-      taskRunId: "taskrun-1",
-      workItemId: "frontend-filter",
-      artifactIds: ["artifact-1"],
+    expect(requests[0]?.body.userInput).toBe("Fix the filter UI");
+    expect(requests[0]?.body.runtimeContext.taskRun).toBeUndefined();
+    expect(requests[0]?.body.runtimeContext.subjectRefs).toContainEqual({
+      kind: "table-row",
+      id: "row-b2e06f801f-01TASK",
+      role: "task",
     });
   });
 
-  test("records task-run completion after an HTTP agent run returns a dialog", async () => {
+  test("does not write task-context completion after an HTTP agent run returns a dialog", async () => {
     const output = new CaptureOutput();
-    const writebacks: any[] = [];
 
     const result = await runAgentTurn({
       agentName: "frontend-implementer",
@@ -173,12 +171,8 @@ describe("cli agent run client", () => {
       env: { AUTH_TOKEN: "token-123" },
       output,
       runtimeMode: "server",
-      taskRunContext: {
+      taskRowContext: {
         rowDbKey: "row-b2e06f801f-01TASK",
-        workItemId: "frontend-filter",
-      },
-      taskRunRecorder: async (args) => {
-        writebacks.push(args);
       },
       fetchImpl: async () => {
         return Response.json({ content: "patched", dialogId: "dialog-1" });
@@ -186,16 +180,6 @@ describe("cli agent run client", () => {
     });
 
     expect(result).toEqual({ exitCode: 0, dialogId: "dialog-1" });
-    expect(writebacks).toHaveLength(1);
-    expect(writebacks[0]).toMatchObject({
-      status: "completed",
-      dialogId: "dialog-1",
-      resultSummary: "patched",
-    });
-    expect(writebacks[0].options.taskRunContext).toEqual({
-      rowDbKey: "row-b2e06f801f-01TASK",
-      workItemId: "frontend-filter",
-    });
   });
 
   test("sends image inputs as multimodal userInput over HTTP", async () => {
@@ -237,6 +221,7 @@ describe("cli agent run client", () => {
       category: "manual-checks",
       inheritedFromDialogKey: "dialog-user-1-dialog-2",
       parentDialogId: "dialog-2",
+      subjectDialogKey: "01SUBJECTDIALOG",
       background: true,
       noStream: true,
       timeoutMs: 600000,
@@ -261,12 +246,45 @@ describe("cli agent run client", () => {
       timeoutMs: 600000,
       stream: false,
     });
+    expect(requests[0]?.body.runtimeContext.subjectRefs).toEqual([
+      { kind: "dialog", id: "01SUBJECTDIALOG", role: "subject" },
+    ]);
     expect(result).toEqual({ exitCode: 0, dialogId: "dialog-1" });
   });
 
-  test("records background task-run handoff as running instead of completed", async () => {
+  test("sends explicit subject refs over HTTP for review evidence", async () => {
+    const output = new CaptureOutput();
+    const requests: Array<{ body: any }> = [];
+
+    await runAgentTurn({
+      agentName: "reviewer",
+      agentKey: "agent-reviewer",
+      serverUrl: "https://nolo.chat",
+      message: "Review this implementation",
+      subjectRefs: [
+        { kind: "table-row", id: "row-user-board-task", role: "subject" },
+        { kind: "dialog", id: "dialog-impl", role: "review-target" },
+        { kind: "external", id: "commit:abc123", role: "commit" },
+      ],
+      scriptDir: "C:/missing/scripts",
+      env: { AUTH_TOKEN: "token-123" },
+      output,
+      runtimeMode: "server",
+      fetchImpl: async (_url, init) => {
+        requests.push({ body: JSON.parse(String(init?.body)) });
+        return Response.json({ content: "queued", dialogId: "dialog-review" });
+      },
+    });
+
+    expect(requests[0]?.body.runtimeContext.subjectRefs).toEqual([
+      { kind: "table-row", id: "row-user-board-task", role: "subject" },
+      { kind: "dialog", id: "dialog-impl", role: "review-target" },
+      { kind: "external", id: "commit:abc123", role: "commit" },
+    ]);
+  });
+
+  test("does not write background handoff state", async () => {
     const output = new CaptureOutput();
-    const writebacks: any[] = [];
 
     const result = await runAgentTurn({
       agentName: "frontend-implementer",
@@ -278,12 +296,8 @@ describe("cli agent run client", () => {
       runtimeMode: "server",
       background: true,
       noStream: true,
-      taskRunContext: {
+      taskRowContext: {
         rowDbKey: "row-b2e06f801f-01TASK",
-        workItemId: "self",
-      },
-      taskRunRecorder: async (args) => {
-        writebacks.push(args);
       },
       output,
       fetchImpl: async () => {
@@ -292,11 +306,6 @@ describe("cli agent run client", () => {
     });
 
     expect(result).toEqual({ exitCode: 0, dialogId: "dialog-bg" });
-    expect(writebacks).toHaveLength(1);
-    expect(writebacks[0]).toMatchObject({
-      status: "running",
-      dialogId: "dialog-bg",
-    });
   });
 
   test("runs forced local turns through the injected runtime adapter without HTTP", async () => {
@@ -571,12 +580,12 @@ describe("cli agent run client", () => {
       round: 1,
       tool: "execShell",
       summary: expect.stringContaining("exit=0"),
+      metadata: { exitCode: 0 },
     });
   });
 
-  test("passes local tool evidence to task-run writeback", async () => {
+  test("does not pass local tool evidence to an external writeback path", async () => {
     const output = new CaptureOutput();
-    const writebacks: any[] = [];
     let completeCalls = 0;
 
     const result = await runAgentTurn({
@@ -588,12 +597,8 @@ describe("cli agent run client", () => {
       env: { AUTH_TOKEN: "token-123" },
       output,
       runtimeMode: "local",
-      taskRunContext: {
+      taskRowContext: {
         rowDbKey: "row-b2e06f801f-01TASK",
-        workItemId: "self",
-      },
-      taskRunRecorder: async (args) => {
-        writebacks.push(args);
       },
       localRuntimeAdapter: {
         host: "cli",
@@ -630,21 +635,10 @@ describe("cli agent run client", () => {
     });
 
     expect(result).toMatchObject({ exitCode: 0, dialogId: "dialog-local" });
-    expect(writebacks).toHaveLength(1);
-    expect(writebacks[0].toolEvidence).toEqual({
-      toolsUsed: ["readFile"],
-      summary: {
-        callCount: 1,
-        resultCount: 1,
-        errorCount: 0,
-        rounds: 1,
-      },
-    });
   });
 
-  test("records local background task-run handoff as running instead of completed", async () => {
+  test("does not write local background handoff state", async () => {
     const output = new CaptureOutput();
-    const writebacks: any[] = [];
 
     const result = await runAgentTurn({
       agentName: "frontend",
@@ -656,12 +650,8 @@ describe("cli agent run client", () => {
       output,
       runtimeMode: "local",
       background: true,
-      taskRunContext: {
+      taskRowContext: {
         rowDbKey: "row-b2e06f801f-01TASK",
-        workItemId: "self",
-      },
-      taskRunRecorder: async (args) => {
-        writebacks.push(args);
       },
       localRuntimeAdapter: {
         host: "cli",
@@ -685,12 +675,6 @@ describe("cli agent run client", () => {
     });
 
     expect(result).toEqual({ exitCode: 0, dialogId: "dialog-local-bg" });
-    expect(writebacks).toHaveLength(1);
-    expect(writebacks[0]).toMatchObject({
-      status: "running",
-      dialogId: "dialog-local-bg",
-    });
-    expect(writebacks[0].resultSummary).toBeUndefined();
   });
 
   test("auto mode prefers a working local runtime before HTTP", async () => {
@@ -910,7 +894,7 @@ describe("cli agent run client", () => {
           prompt: "Implement frontend tasks.",
           provider: "agy",
           cliProvider: "agy",
-          toolNames: ["streamParallelAgents", "taskRun"],
+          toolNames: ["streamParallelAgents", "queryTableRows"],
         }),
         loadDialogHistory: async () => [],
         saveTurn: async () => ({ dialogId: "dialog-agy-auto-local" }),
@@ -1116,7 +1100,7 @@ describe("cli agent run client", () => {
           model: "fake-local",
           runtimeToolPolicy: {
             version: 1,
-            agentTools: ["queryTableRows", "taskRun"],
+            agentTools: ["queryTableRows", "streamParallelAgents"],
             runtimeTools: ["execShell"],
           },
         }),
@@ -1141,7 +1125,7 @@ describe("cli agent run client", () => {
     expect(httpCalls).toHaveLength(1);
     expect(httpCalls[0]?.body.agentKey).toBe("agent-policy-platform-tools");
     expect(output.text()).toContain("auto runtime: skipping local runtime");
-    expect(output.text()).toContain("queryTableRows, taskRun");
+    expect(output.text()).toContain("queryTableRows, streamParallelAgents");
   });
 
   test("auto mode skips known platform agents when local config cannot be read", async () => {