numerous fixes and tests

Author: v1r3n

.claude/plans/tranquil-sleeping-leaf.md

@@ -0,0 +1,36 @@
+# Plan: Fix 8 Failing Integration Tests in WorkflowExecutor.test.ts
+
+## Context
+
+Integration tests ran: **20 passed, 2 failed** (286/295 tests pass). The 8 failures are all in the "Execute with Return Strategy and Consistency" describe block of `WorkflowExecutor.test.ts`. The 1 `ServiceRegistryClient.test.ts` failure is deferred.
+
+**Root cause:** The 3 complex workflow definitions use HTTP tasks calling `http://httpbin:8081/api/hello?name=test1`. The `httpbin` hostname is a Docker service only available in Docker Compose environments. Locally, DNS fails → HTTP tasks fail → workflows FAIL → YIELD tasks never reached → signals can't work → `responseType` is `undefined`.
+
+## Fix
+
+Change `optional: false` → `optional: true` on the HTTP task in each of the 3 workflow metadata files. This lets workflows continue past failed HTTP tasks to reach YIELD tasks. The tests only care about signal/return strategy behavior, not HTTP results.
+
+### Files to modify (1 line each)
+
+1. **`src/integration-tests/metadata/complex_wf_signal_test.ts`** line 25
+2. **`src/integration-tests/metadata/complex_wf_signal_test_subworkflow_1.ts`** line 25
+3. **`src/integration-tests/metadata/complex_wf_signal_test_subworkflow_2.ts`** line 25
+
+Only the HTTP tasks get `optional: true`. YIELD and SUB_WORKFLOW tasks remain `optional: false`.
+
+Add a comment: `// optional so tests work without httpbin Docker service`
+
+## Why this is safe
+
+- When httpbin IS available (CI/Docker): HTTP task succeeds normally, `optional: true` has no effect
+- When httpbin is NOT available (local): HTTP task fails gracefully, workflow continues
+- No other tests reference these metadata files
+- Test assertions check workflow/task fields that are still populated regardless of HTTP task status
+
+## Verification
+
+```bash
+npm run test:integration:orkes-v5 -- --testPathPatterns=WorkflowExecutor
+```
+
+All 8 previously-failing tests should pass. Then re-run full suite to confirm no regressions.

.env.example

@@ -3,4 +3,7 @@ CONDUCTOR_SERVER_URL=""
 CONDUCTOR_AUTH_KEY=""
 CONDUCTOR_AUTH_SECRET=""
 
-CONDUCTOR_MAX_HTTP2_CONNECTIONS=
+CONDUCTOR_MAX_HTTP2_CONNECTIONS=
+
+CONDUCTOR_TLS_INSECURE=
+CONDUCTOR_DISABLE_HTTP2=

AGENTS.md

@@ -45,7 +45,7 @@ src/integration-tests/           # E2E tests against real Conductor server
 ## Commands
 
 ```bash
-npm test                           # Unit tests (469 tests)
+npm test                           # Unit tests (482+ tests)
 npm run build                      # tsup (ESM + CJS dual output)
 npm run lint                       # ESLint
 npm run generate-openapi-layer     # Regenerate from OpenAPI spec
@@ -61,12 +61,12 @@ npm run test:integration:orkes-v5
 
 After every code change, you **must** run the following before considering the work complete:
 
-1. **Lint** — fix all lint errors in files you changed:
+1. **Lint** — **MANDATORY at the end of every session**. Run the linter and fix ALL errors before finishing:
    ```bash
-   npm run lint                          # Check all files
+   npm run lint                          # Check all files — must show 0 errors
    npx eslint --fix src/path/to/file.ts  # Auto-fix a specific file
    ```
-   You must fix all lint errors in files you modified. Do not introduce new lint violations.
+   You must fix all lint errors in files you modified. Do not introduce new lint violations. **Do not end a session with lint errors remaining.**
 
 2. **Unit tests** — all must pass:
    ```bash
@@ -232,4 +232,4 @@ When adding new client methods, builders, worker features, or examples:
 | ConductorWorkflow DSL | Yes | 100% | |
 | Worker + Metrics + Context | Yes | 100% | |
 | Error paths | - | 35 tests | |
-| **Total** | **469** | **191** | **660** |
+| **Total** | **482+** | **191** | **673+** |

SDK_NEW_LANGUAGE_GUIDE.md

@@ -110,7 +110,12 @@ async function main() {
   }
 
   const humanTaskEntry = pendingTasks[0];
-  const humanTaskId = humanTaskEntry.taskId!;
+  const humanTaskId = humanTaskEntry?.taskId;
+  if (!humanTaskId) {
+    console.log("  Human task has no taskId. Exiting.");
+    await handler.stopWorkers();
+    process.exit(1);
+  }
   console.log(`  Found human task: ${humanTaskId}`);
   console.log(`  State: ${humanTaskEntry.state}`);
   console.log(`  Display name: ${humanTaskEntry.humanTaskDef?.displayName}\n`);

examples/advanced/human-tasks.ts

@@ -26,7 +26,8 @@ async function main() {
 
     // Create application
     const app = await appClient.createApplication(appName);
-    applicationId = app.id!;
+    if (!app.id) throw new Error("Expected application id");
+    applicationId = app.id;
     console.log(`1. Created application: ${app.name} (id: ${applicationId})`);
 
     // Get application

examples/api-journeys/applications.ts

@@ -27,6 +27,7 @@
       "name": "James Stuart Milne"
     }
   ],
+  "sideEffects": false,
   "engines": {
     "node": ">=18"
   },

package.json

@@ -8,15 +8,13 @@ import {
 } from "@jest/globals";
 import {
   TaskType,
-  type WorkflowDef,
   type Client,
 } from "../open-api";
 import {
   ConductorWorkflow,
   MetadataClient,
   WorkflowExecutor,
   orkesConductorClient,
-  simpleTask,
   setVariableTask,
   llmChatCompleteTask,
   Role,

src/integration-tests/ConductorWorkflow.test.ts

@@ -47,7 +47,7 @@ describe("E2E: 5-task workflow × 50 executions", () => {
 
         worker({ taskDefName: taskName, pollInterval: 100, concurrency: 5 })(
           async function taskWorker(task: Task) {
-            executionCounts[taskName]!++;
+            executionCounts[taskName] = (executionCounts[taskName] ?? 0) + 1;
             return {
               status: "COMPLETED" as const,
               outputData: {
@@ -121,15 +121,19 @@ describe("E2E: 5-task workflow × 50 executions", () => {
       expect(results.length).toBe(WORKFLOW_COUNT);
 
       for (let w = 0; w < results.length; w++) {
-        const wf = results[w]!;
+        const wf = results[w];
+        if (!wf) throw new Error(`Expected result at index ${w}`);
         expect(wf.status).toBe("COMPLETED");
 
         // Each workflow should have exactly 5 tasks
         expect(wf.tasks?.length).toBe(TASK_COUNT);
 
         // Validate each task's output
         for (let t = 0; t < TASK_COUNT; t++) {
-          const task = wf.tasks![t]!;
+          const tasks = wf.tasks;
+          if (!tasks) throw new Error(`Expected tasks for workflow at index ${w}`);
+          const task = tasks[t];
+          if (!task) throw new Error(`Expected task at index ${t}`);
           expect(task.status).toBe("COMPLETED");
           expect(task.outputData?.taskNumber).toBe(t + 1);
           expect(task.outputData?.message).toBe(

src/integration-tests/E2EFiveTaskWorkflow.test.ts

@@ -1,4 +1,4 @@
-import { expect, describe, test, jest, beforeAll, beforeEach, afterAll } from "@jest/globals";
+import { expect, describe, test, jest, beforeAll, afterAll } from "@jest/globals";
 import {
   orkesConductorClient,
   OrkesClients,