@zibby/core 0.3.6 → 0.3.7

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (40) hide show
  1. package/dist/index.js +34 -32
  2. package/dist/package.json +2 -1
  3. package/dist/register-built-in-strategies.js +25 -23
  4. package/dist/strategies/claude-strategy.js +3 -1
  5. package/dist/strategies/index.js +25 -23
  6. package/dist/templates/browser-test-automation/graph.mjs +20 -6
  7. package/dist/templates/browser-test-automation/nodes/preflight.mjs +50 -15
  8. package/dist/templates/browser-test-automation/state.js +61 -0
  9. package/dist/templates/code-analysis/README.md +60 -0
  10. package/dist/templates/code-analysis/graph.mjs +33 -0
  11. package/dist/templates/code-analysis/nodes/analyze-ticket-node.js +1 -1
  12. package/dist/templates/code-analysis/nodes/create-pr-node.js +1 -1
  13. package/dist/templates/code-analysis/nodes/generate-code-node.js +1 -1
  14. package/dist/templates/code-analysis/nodes/generate-test-cases-node.js +1 -1
  15. package/dist/templates/code-analysis/nodes/services/prMetaService.js +1 -1
  16. package/dist/templates/code-analysis/state.js +14 -6
  17. package/dist/templates/generate-test-cases/README.md +72 -0
  18. package/dist/templates/generate-test-cases/graph.mjs +46 -0
  19. package/dist/templates/generate-test-cases/nodes/generate-test-cases-node.js +381 -0
  20. package/dist/templates/generate-test-cases/nodes/setup-node.js +142 -0
  21. package/dist/templates/generate-test-cases/state.js +54 -0
  22. package/dist/templates/index.js +53 -0
  23. package/package.json +2 -1
  24. package/templates/browser-test-automation/graph.mjs +20 -6
  25. package/templates/browser-test-automation/nodes/preflight.mjs +50 -15
  26. package/templates/browser-test-automation/state.js +61 -0
  27. package/templates/code-analysis/README.md +60 -0
  28. package/templates/code-analysis/graph.mjs +33 -0
  29. package/templates/code-analysis/nodes/analyze-ticket-node.js +1 -1
  30. package/templates/code-analysis/nodes/create-pr-node.js +1 -1
  31. package/templates/code-analysis/nodes/generate-code-node.js +1 -1
  32. package/templates/code-analysis/nodes/generate-test-cases-node.js +1 -1
  33. package/templates/code-analysis/nodes/services/prMetaService.js +1 -1
  34. package/templates/code-analysis/state.js +14 -6
  35. package/templates/generate-test-cases/README.md +72 -0
  36. package/templates/generate-test-cases/graph.mjs +46 -0
  37. package/templates/generate-test-cases/nodes/generate-test-cases-node.js +381 -0
  38. package/templates/generate-test-cases/nodes/setup-node.js +142 -0
  39. package/templates/generate-test-cases/state.js +54 -0
  40. package/templates/index.js +53 -0
@@ -0,0 +1,33 @@
1
+ /**
2
+ * code-analysis template — scaffoldable WorkflowAgent wrapper.
3
+ *
4
+ * Why this file exists alongside graph.js:
5
+ * - graph.js exports `buildAnalysisGraph(graph)` — a builder
6
+ * function the existing `zibby analyze-graph` cli + backend
7
+ * analysis handlers consume directly. Its shape can't change
8
+ * without breaking those callers.
9
+ * - To make code-analysis scaffoldable via
10
+ * `zibby workflow new <slug> -t code-analysis` and runnable via
11
+ * `zibby workflow run <slug>` / `zibby workflow deploy <slug>`,
12
+ * the template needs to expose a WorkflowAgent class — same
13
+ * contract as browser-test-automation.
14
+ *
15
+ * This .mjs is the entry the user-facing scaffold uses; graph.js
16
+ * stays put for internal callers.
17
+ */
18
+
19
+ import { WorkflowAgent, WorkflowGraph } from '@zibby/core';
20
+ import { buildAnalysisGraph } from './graph.js';
21
+
22
+ export class CodeAnalysisAgent extends WorkflowAgent {
23
+ buildGraph() {
24
+ const graph = new WorkflowGraph();
25
+ buildAnalysisGraph(graph);
26
+ return graph;
27
+ }
28
+
29
+ async onComplete(result) {
30
+ const ok = result.success !== false;
31
+ console.log(`[code-analysis] complete — success: ${ok}`);
32
+ }
33
+ }
@@ -2,7 +2,7 @@ import { existsSync, readFileSync } from 'fs';
2
2
  import { join } from 'path';
3
3
  import { randomBytes } from 'crypto';
4
4
  import { z } from 'zod';
5
- import { adfToText } from '../../../src/utils/adf-converter.js';
5
+ import { adfToText } from '@zibby/core/utils/adf-converter.js';
6
6
 
7
7
  const generateId = () => randomBytes(16).toString('hex');
8
8
 
@@ -1,5 +1,5 @@
1
1
  import axios from 'axios';
2
- import { adfToText } from '../../../src/utils/adf-converter.js';
2
+ import { adfToText } from '@zibby/core/utils/adf-converter.js';
3
3
  import { z } from 'zod';
4
4
 
5
5
  const CreatePROutputSchema = z.object({
@@ -10,7 +10,7 @@ import { existsSync, readFileSync } from 'fs';
10
10
  import Handlebars from 'handlebars';
11
11
  import { invokeAgent } from '@zibby/core';
12
12
  import { generatePRMeta } from './services/prMetaService.js';
13
- import { adfToText } from '../../../src/utils/adf-converter.js';
13
+ import { adfToText } from '@zibby/core/utils/adf-converter.js';
14
14
  import { z } from 'zod';
15
15
 
16
16
  const CodeImplementationOutputSchema = z.object({
@@ -11,7 +11,7 @@
11
11
  import { invokeAgent } from '@zibby/core';
12
12
  import { z } from 'zod';
13
13
  import { randomBytes } from 'crypto';
14
- import { adfToText } from '../../../src/utils/adf-converter.js';
14
+ import { adfToText } from '@zibby/core/utils/adf-converter.js';
15
15
 
16
16
  // Generate a simple unique ID
17
17
  const generateId = () => randomBytes(8).toString('hex');
@@ -4,7 +4,7 @@
4
4
 
5
5
  import { z } from 'zod';
6
6
  import { invokeAgent } from '@zibby/core';
7
- import { adfToText } from '../../../../src/utils/adf-converter.js';
7
+ import { adfToText } from '@zibby/core/utils/adf-converter.js';
8
8
 
9
9
  const PRMetaSchema = z.object({
10
10
  prTitle: z.string().describe('Short PR title that includes the ticket key'),
@@ -31,10 +31,18 @@ export const analysisStateSchema = z.object({
31
31
  githubToken: z.string().optional().describe('GitHub personal access token'),
32
32
  model: z.string().default('auto').describe('AI model to use'),
33
33
  nodeConfigs: z.record(z.string(), z.any()).optional().describe('Per-node configuration overrides'),
34
-
35
- EXECUTION_ID: z.string().nullable().optional(),
36
- PROGRESS_QUEUE_URL: z.string().url().nullable().optional(),
37
- PROGRESS_API_URL: z.string().url().nullable().optional(),
38
- SQS_AUTH_TOKEN: z.string().nullable().optional(),
39
- PROJECT_API_TOKEN: z.string().nullable().optional(),
40
34
  });
35
+
36
+ // NOTE: legacy `EXECUTION_ID`, `PROGRESS_QUEUE_URL`, `PROGRESS_API_URL`,
37
+ // `SQS_AUTH_TOKEN`, `PROJECT_API_TOKEN` fields are no longer declared in
38
+ // this schema. They were used only by the legacy `analyze-graph` cli +
39
+ // the analysis UI's progress reporter — never by template nodes. Zod's
40
+ // safeParse here doesn't strip unknown keys (see graph.js:504), so the
41
+ // legacy flow still works at runtime: analyze-graph.js injects them
42
+ // into initialState, validation passes, nodes ignore them, the
43
+ // progress-reporter middleware reads them off the state argument.
44
+ //
45
+ // Keeping them out of the schema means scaffolded user copies (via
46
+ // `zibby workflow new -t code-analysis`) ship a clean state that
47
+ // reflects only what the template actually needs — no false coupling
48
+ // to internal infra fields.
@@ -0,0 +1,72 @@
1
+ # Generate Test Cases Template
2
+
3
+ Standalone slice of `code-analysis` — takes a code diff you already
4
+ have and generates plain-English test specifications for it. Skips
5
+ ticket-analysis and code-gen.
6
+
7
+ ## When to use
8
+
9
+ - You have a PR diff (or `git diff` output) and want test specs for the changes.
10
+ - You don't need the LLM to generate the code itself — that already exists.
11
+ - You want a smaller, faster pipeline than the full code-analysis workflow.
12
+
13
+ If you want the full pipeline (analyze ticket → generate code → generate
14
+ tests in one shot), use `code-analysis` instead.
15
+
16
+ ## Graph
17
+
18
+ ```
19
+ setup ─→ generate_test_cases
20
+ ```
21
+
22
+ - `setup` — clone repos so the LLM can explore actual routing /
23
+ components when writing test steps
24
+ - `generate_test_cases` — LLM reads the diff + ticket context, emits
25
+ 4–8 plain-English test specs (with priority + category)
26
+
27
+ ## Required state inputs
28
+
29
+ ```js
30
+ {
31
+ workspace: '/abs/path/to/workspace',
32
+ repos: [{
33
+ name: 'my-app',
34
+ url: 'https://github.com/org/my-app.git',
35
+ branch: 'main',
36
+ isPrimary: true,
37
+ }],
38
+ ticketContext: {
39
+ ticketKey: 'PROJ-123',
40
+ summary: 'short title',
41
+ description: 'long description',
42
+ },
43
+ codeImplementation: {
44
+ diff: '...unified diff string...',
45
+ changedFiles: ['src/auth/login.ts', 'src/components/LoginForm.tsx'],
46
+ },
47
+ githubToken: process.env.GITHUB_TOKEN,
48
+ }
49
+ ```
50
+
51
+ The `codeImplementation` field is what makes this template standalone —
52
+ in `code-analysis` it's produced by the upstream `generate_code` node;
53
+ here you supply it directly.
54
+
55
+ See `state.js` for the full Zod schema.
56
+
57
+ ## Customizing the prompt
58
+
59
+ The test-generation prompt is inlined in
60
+ `nodes/generate-test-cases-node.js` (the `buildTestGenerationPrompt`
61
+ function). Edit that to change the LLM's instructions — what
62
+ exploration steps it follows, what test format it returns, etc.
63
+
64
+ ## Cloud deployment
65
+
66
+ ```bash
67
+ zibby workflow deploy <your-slug>
68
+ zibby workflow trigger <uuid> --input '{"workspace": "...", "codeImplementation": {...}, ...}'
69
+ ```
70
+
71
+ Cloud runs need network access for `git clone` (the `setup` node
72
+ shells out to `git`). Default cloud egress works.
@@ -0,0 +1,46 @@
1
+ /**
2
+ * generate-test-cases template — standalone slice of code-analysis.
3
+ *
4
+ * Two-node graph:
5
+ *
6
+ * setup ─→ generate_test_cases
7
+ *
8
+ * Use case: "I have a PR diff already; generate test cases for it."
9
+ * Skips ticket-analysis + code-gen — the user provides the diff
10
+ * directly as `state.codeImplementation`.
11
+ *
12
+ * Why duplicated nodes/setup-node.js + nodes/generate-test-cases-node.js
13
+ * (rather than importing from `../code-analysis/nodes/...`):
14
+ * Templates have to be self-contained — when scaffolded into a user's
15
+ * project at `.zibby/workflows/<slug>/`, there's no sibling
16
+ * `code-analysis/` folder to import from. browser-test + code-analysis
17
+ * follow the same convention. The cost is keeping the duplicates in
18
+ * sync; the win is each template is independently editable.
19
+ */
20
+
21
+ import { WorkflowAgent, WorkflowGraph } from '@zibby/core';
22
+ import { setupNode } from './nodes/setup-node.js';
23
+ import { generateTestCasesNode } from './nodes/generate-test-cases-node.js';
24
+ import { generateTestCasesStateSchema } from './state.js';
25
+
26
+ export class GenerateTestCasesAgent extends WorkflowAgent {
27
+ buildGraph() {
28
+ const graph = new WorkflowGraph();
29
+ graph.setStateSchema(generateTestCasesStateSchema);
30
+
31
+ graph.addNode('setup', setupNode);
32
+ graph.addNode('generate_test_cases', generateTestCasesNode);
33
+
34
+ graph.setEntryPoint('setup');
35
+ graph.addEdge('setup', 'generate_test_cases');
36
+ graph.addEdge('generate_test_cases', 'END');
37
+
38
+ return graph;
39
+ }
40
+
41
+ async onComplete(result) {
42
+ const tests = result.state?.generate_test_cases?.tests;
43
+ const count = tests?.structured?.testCases?.length || 0;
44
+ console.log(`[generate-test-cases] complete — ${count} test case(s) generated`);
45
+ }
46
+ }
@@ -0,0 +1,381 @@
1
+ /**
2
+ * Generate Test Cases Node - Generate human-readable test case specifications
3
+ * Used by: analysisGraph, implementationGraph
4
+ *
5
+ * Output: Plain-text test specifications for AI interpretation
6
+ * - High-level test steps in natural language
7
+ * - Includes credentials, prerequisites, and expected results
8
+ * - Can be executed by AI agents (like MCP) and converted to Playwright
9
+ */
10
+
11
+ import { invokeAgent } from '@zibby/core';
12
+ import { z } from 'zod';
13
+ import { randomBytes } from 'crypto';
14
+ import { adfToText } from '@zibby/core/utils/adf-converter.js';
15
+
16
+ // Generate a simple unique ID
17
+ const generateId = () => randomBytes(8).toString('hex');
18
+
19
+ // Schema for plain-text test specification
20
+ const TestSpecificationSchema = z.object({
21
+ id: z.string().optional(),
22
+ title: z.string().describe('Test name (e.g., "Login Functionality", "Create New User")'),
23
+ application: z.string().describe('Application name being tested'),
24
+ url: z.string().describe('Base URL of the application'),
25
+ testerRole: z.string().describe('Role performing the test (e.g., "Admin user", "End user")'),
26
+ feature: z.string().describe('Feature being tested (e.g., "Authentication - Login")'),
27
+ ticketId: z.string().describe('Associated ticket ID'),
28
+ testCredentials: z.array(z.object({
29
+ field: z.string().describe('Credential field name (e.g., "Email", "Password", "API Key")'),
30
+ value: z.string().describe('Credential value')
31
+ })).optional().describe('Test credentials needed (if any)'),
32
+ testObjective: z.string().describe('One clear sentence: what this test verifies'),
33
+ testSteps: z.array(z.string()).describe('Plain English steps - natural language instructions (e.g., "Navigate to /login", "Enter email: test@example.com", "Click the submit button")'),
34
+ expectedResults: z.array(z.string()).describe('What should happen - bullet points of expected outcomes'),
35
+ status: z.string().default('Ready for automation').describe('Test status'),
36
+ priority: z.enum(['Critical', 'High', 'Medium', 'Low']).describe('Test priority'),
37
+ category: z.string().describe('Test category (e.g., "Smoke Test", "Regression", "Integration")')
38
+ });
39
+
40
+ const CategoryCountSchema = z.object({
41
+ category: z.string().describe('Category name'),
42
+ count: z.number().describe('Number of tests in this category')
43
+ });
44
+
45
+ // Schema for AI agent output (what invokeAgent validates)
46
+ const TestSuiteSchema = z.object({
47
+ testCases: z.array(TestSpecificationSchema).describe('Array of test specifications'),
48
+ summary: z.object({
49
+ totalTests: z.number(),
50
+ byCategory: z.array(CategoryCountSchema).optional(),
51
+ coverageNotes: z.string().optional()
52
+ }).optional()
53
+ });
54
+
55
+ // Node output schema (what this node returns to state)
56
+ const GenerateTestCasesOutputSchema = z.object({
57
+ success: z.boolean(),
58
+ tests: z.object({
59
+ structured: z.object({
60
+ testCases: z.array(z.any()),
61
+ summary: z.object({
62
+ totalTests: z.number(),
63
+ byCategory: z.array(CategoryCountSchema).optional()
64
+ }).optional()
65
+ }),
66
+ timestamp: z.string()
67
+ }).nullable()
68
+ });
69
+
70
+ export const generateTestCasesNode = {
71
+ name: 'generate_test_cases',
72
+
73
+ // Output schema defines what this node returns to state
74
+ outputSchema: GenerateTestCasesOutputSchema,
75
+
76
+ execute: async (state) => {
77
+ console.log('\n🧪 Generating test cases...');
78
+
79
+ const { ticketContext, workspace, model, nodeConfigs = {} } = state;
80
+ const aiModel = model || ticketContext.model || 'auto';
81
+ // Standalone-template variant: this template runs WITHOUT an upstream
82
+ // generate_code node, so the diff comes in directly as
83
+ // `state.codeImplementation`. Keep the original two fallbacks so the
84
+ // node still works if someone wires it after a code-gen node.
85
+ const codeImpl = state.codeImplementation ||
86
+ state.generate_code?.codeImplementation ||
87
+ state.implement_code?.codeImplementation;
88
+
89
+ if (!codeImpl || !codeImpl.diff) {
90
+ console.log('⚠️ No code changes detected, skipping test generation');
91
+ return {
92
+ success: true,
93
+ tests: null
94
+ };
95
+ }
96
+
97
+ // Get node config for this node (if user configured it)
98
+ const nodeConfig = nodeConfigs.generate_test_cases || {};
99
+ console.log('📋 Node config:', JSON.stringify(nodeConfig, null, 2));
100
+
101
+ // Build test generation prompt with workspace path for codebase exploration
102
+ const testPrompt = buildTestGenerationPrompt(ticketContext, codeImpl, workspace, nodeConfig);
103
+
104
+ console.log(`🚀 Running AI Agent to generate tests with model: ${aiModel}...`);
105
+
106
+ const agentResult = await invokeAgent(testPrompt, {
107
+ state,
108
+ model: aiModel,
109
+ schema: TestSuiteSchema
110
+ });
111
+
112
+ const validatedOutput = agentResult?.structured || agentResult;
113
+
114
+ // Convert structured AI output → simple { id, title, priority, content } format
115
+ const testCases = (validatedOutput?.testCases || []).map(test => ({
116
+ id: generateId(),
117
+ title: test.title,
118
+ priority: test.priority || 'Medium',
119
+ content: renderTestCaseToText(test),
120
+ status: 'pending'
121
+ }));
122
+
123
+ console.log(`✅ Generated ${testCases.length} test cases`);
124
+
125
+ return {
126
+ success: true,
127
+ tests: {
128
+ structured: {
129
+ testCases,
130
+ summary: validatedOutput.summary || {
131
+ totalTests: testCases.length,
132
+ byCategory: Object.entries(
133
+ testCases.reduce((acc, tc) => {
134
+ acc[tc.category] = (acc[tc.category] || 0) + 1;
135
+ return acc;
136
+ }, {})
137
+ ).map(([category, count]) => ({ category, count }))
138
+ }
139
+ },
140
+ timestamp: new Date().toISOString()
141
+ }
142
+ };
143
+ }
144
+ };
145
+
146
+ function renderTestCaseToText(test) {
147
+ const lines = [];
148
+
149
+ if (test.testObjective) {
150
+ lines.push('WHAT THIS TESTS');
151
+ lines.push(test.testObjective);
152
+ lines.push('');
153
+ }
154
+
155
+ if (test.url) {
156
+ lines.push('START URL');
157
+ lines.push(test.url);
158
+ lines.push('');
159
+ }
160
+
161
+ if (test.testCredentials && test.testCredentials.length > 0) {
162
+ lines.push('TEST CREDENTIALS');
163
+ test.testCredentials.forEach(cred => {
164
+ lines.push(`• ${cred.field}: ${cred.value}`);
165
+ });
166
+ lines.push('');
167
+ }
168
+
169
+ if (test.testSteps && test.testSteps.length > 0) {
170
+ lines.push('TEST STEPS');
171
+ test.testSteps.forEach((step, i) => {
172
+ lines.push(`${i + 1}. ${step}`);
173
+ });
174
+ lines.push('');
175
+ }
176
+
177
+ if (test.expectedResults && test.expectedResults.length > 0) {
178
+ lines.push('EXPECTED RESULTS');
179
+ test.expectedResults.forEach(result => {
180
+ lines.push(`• ${result}`);
181
+ });
182
+ }
183
+
184
+ return lines.join('\n').trim();
185
+ }
186
+
187
+ function buildTestGenerationPrompt(ticketContext, codeImpl, workspace, nodeConfig = {}) {
188
+ // Extract user-provided context (test URL, admin credentials, etc.)
189
+ const extractContext = nodeConfig.extractContext || {};
190
+ let contextSection = '';
191
+ if (extractContext.testBaseUrl || extractContext.adminUsername || extractContext.testAccountUsername) {
192
+ contextSection = `
193
+ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
194
+ PROVIDED TEST CONTEXT (USE THESE IN TEST CASES)
195
+ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
196
+
197
+ ${extractContext.testBaseUrl ? `Test Base URL: ${extractContext.testBaseUrl}` : ''}
198
+ ${extractContext.adminUsername ? `Admin Username: ${extractContext.adminUsername}` : ''}
199
+ ${extractContext.adminPassword ? `Admin Password: ${extractContext.adminPassword}` : ''}
200
+ ${extractContext.testAccountUsername ? `Test Account Username: ${extractContext.testAccountUsername}` : ''}
201
+ ${extractContext.testAccountPassword ? `Test Account Password: ${extractContext.testAccountPassword}` : ''}
202
+
203
+ **IMPORTANT**: Use these exact values in your test cases. Do NOT make up URLs or credentials.
204
+
205
+ `;
206
+ }
207
+
208
+ return `You are an AI test automation engineer creating human-readable test specifications.
209
+
210
+ WORKSPACE: ${workspace}
211
+
212
+ TICKET: ${ticketContext.ticketKey} - ${ticketContext.summary}
213
+ ${ticketContext.description ? `Description: ${typeof ticketContext.description === 'object' ? adfToText(ticketContext.description) : ticketContext.description}` : ''}
214
+
215
+ MODIFIED FILES:
216
+ ${codeImpl.changedFiles.map(f => ` - ${f}`).join('\n')}
217
+
218
+ FULL CODE DIFF:
219
+ \`\`\`diff
220
+ ${codeImpl.diff}
221
+ \`\`\`
222
+
223
+ ${contextSection}━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
224
+
225
+ YOUR JOB: Generate 4-8 test specifications in plain-text format that AI agents can interpret.
226
+
227
+ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
228
+ STEP 1: EXPLORE THE CODEBASE (MANDATORY - DO NOT SKIP)
229
+ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
230
+
231
+ YOU MUST explore the codebase to understand the application structure BEFORE writing tests.
232
+
233
+ REQUIRED EXPLORATION:
234
+
235
+ 1. **Find and Read Routing Files**
236
+ - Search the workspace for routing files: look for common names like App, routes, router, main, index files in .js/.jsx/.ts/.tsx format
237
+ - Check common locations: src/, app/, client/, web/, components/, pages/ directories
238
+ - Identify ALL routes and their URL patterns
239
+ - Pay special attention to routes with parameters (e.g., /stores/:storeId/products, /users/:userId/orders, /items/:itemId)
240
+
241
+ 2. **Understand Route Parameters**
242
+ - Dynamic segments in routes (marked with :paramName or {paramName}) are PLACEHOLDERS
243
+ - You MUST include navigation steps to select/create that resource first
244
+ - Example: To reach /stores/:storeId/products, you must first navigate to /stores list and select a specific store
245
+ - Example: To reach /orders/:orderId/details, you must first navigate to /orders list and click on a specific order
246
+
247
+ 3. **Map Navigation Flow**
248
+ - Understand the hierarchy: authentication → landing/home → resource lists → individual resources → sub-features
249
+ - Find what links, buttons, tabs exist on each page to navigate to the next level
250
+ - Read page components to understand the UI structure and navigation elements
251
+
252
+ 4. **Find Authentication**
253
+ - Check for login/signin routes, auth components, protected routes, authentication guards
254
+ - Look for HOCs, route wrappers, or middleware that handle authentication
255
+ - Understand if authentication is required and what the login flow looks like
256
+
257
+ 5. **Identify Page Components**
258
+ - Find the actual page/view components referenced in routes
259
+ - Read their code to understand imports, props, state management, and structure
260
+ - Understand what data they load, what APIs they call, and what user actions they support
261
+
262
+ CRITICAL RULES:
263
+ - Routes with dynamic parameters (like :id, :userId, :orderId) REQUIRE parent resource selection
264
+ - You CANNOT jump directly to a URL with an ID - you must navigate through the parent list and SELECT that item
265
+ - Base your test steps on ACTUAL navigation paths you discover in the codebase
266
+ - If you don't explore properly, your tests will have WRONG navigation and be unusable by AI agents
267
+
268
+ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
269
+ STEP 2: GENERATE TEST SPECIFICATIONS
270
+ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
271
+
272
+ For each test, provide:
273
+ - **title**: Test name (e.g., "Filter Orders by Status")
274
+ - **application**: Application name (infer from package.json or codebase)
275
+ - **url**: Base application URL (usually http://localhost:3000 or staging URL if found)
276
+ - **testerRole**: Who performs this test (e.g., "Admin user", "End user", "Guest")
277
+ - **feature**: Feature being tested (match the modified files)
278
+ - **ticketId**: Use the ticket key provided above
279
+ - **testCredentials**: Array of credentials needed (if authentication required):
280
+ - field: "Email" / "Password" / "Username" / "API Key", etc.
281
+ - value: Use realistic test data (e.g., "admin@example.com", "TestPass123")
282
+ - **testObjective**: One clear sentence describing what this test verifies
283
+ - **testSteps**: Plain English steps - MUST match actual navigation flow you discovered. Example:
284
+ * "Navigate to login page http://localhost:3000/login"
285
+ * "Enter email: admin@example.com"
286
+ * "Enter password: TestPass123"
287
+ * "Click the Sign In button"
288
+ * "Navigate to Projects page http://localhost:3000/projects"
289
+ * "Click on project 'Test Project 1'"
290
+ * "Navigate to Orders tab"
291
+ * "Click the status filter dropdown"
292
+ * "Select 'Completed' from the filter"
293
+ * "Verify only completed orders are displayed"
294
+ DON'T use technical terms like "action", "target", "selector" - just natural language
295
+ - **expectedResults**: Bullet points of what should happen
296
+ - **priority**: Critical, High, Medium, or Low
297
+ - **category**: Test category (e.g., "Smoke Test", "Regression", "Integration")
298
+
299
+ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
300
+ EXAMPLE: Multi-Level Navigation (Generic Restaurant Ordering App)
301
+ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
302
+
303
+ Assume this routing structure (discovered from App.jsx):
304
+ /login
305
+ /restaurants
306
+ /restaurants/:restaurantId/menu
307
+ /restaurants/:restaurantId/menu/:itemId/details
308
+
309
+ CORRECT Test Steps (follows nested structure):
310
+ {
311
+ "title": "Add Menu Item to Cart with Size Selection",
312
+ "application": "Restaurant Ordering System",
313
+ "url": "http://localhost:3000",
314
+ "testerRole": "Customer",
315
+ "feature": "Menu - Item Size Selection",
316
+ "ticketId": "REST-456",
317
+ "testCredentials": [
318
+ { "field": "Email", "value": "customer@email.com" },
319
+ { "field": "Password", "value": "Order123" }
320
+ ],
321
+ "testObjective": "Verify customers can select different sizes when adding menu items to cart",
322
+ "testSteps": [
323
+ "Navigate to login page http://localhost:3000/login",
324
+ "Enter email: customer@email.com",
325
+ "Enter password: Order123",
326
+ "Click the Sign In button",
327
+ "Navigate to Restaurants page http://localhost:3000/restaurants",
328
+ "Click on restaurant 'Downtown Pizza'",
329
+ "Wait for menu to load",
330
+ "Scroll to 'Beverages' section",
331
+ "Click on menu item 'Lemonade'",
332
+ "Wait for item details to display",
333
+ "Click the size dropdown",
334
+ "Select 'Large' size option",
335
+ "Verify price updates to large size price",
336
+ "Click Add to Cart button",
337
+ "Verify item added to cart with Large size"
338
+ ],
339
+ "expectedResults": [
340
+ "Size dropdown displays all available sizes (Small, Medium, Large)",
341
+ "Price updates when different size is selected",
342
+ "Item is added to cart with correct size",
343
+ "Cart icon shows updated item count",
344
+ "No errors displayed during the process"
345
+ ],
346
+ "priority": "High",
347
+ "category": "Smoke Test"
348
+ }
349
+
350
+ WRONG Test Steps (skips required navigation):
351
+ ❌ "Navigate to Menu page" (Too vague - which restaurant?)
352
+ ❌ "Go to /restaurants/123/menu" (Can't directly use IDs without selection)
353
+ ❌ "Select item from menu" (Need to navigate to restaurant first)
354
+
355
+ CORRECT Test Steps (shows proper navigation):
356
+ ✅ "Navigate to Restaurants page http://localhost:3000/restaurants"
357
+ ✅ "Click on restaurant 'Downtown Pizza'"
358
+ ✅ "Wait for menu to load"
359
+ ✅ "Click on menu item 'Lemonade'"
360
+
361
+ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
362
+ FINAL REMINDERS
363
+ ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
364
+
365
+ 1. ALWAYS explore routing files FIRST
366
+ 2. NEVER skip navigation steps - include full path from login to feature
367
+ 3. If routes have :parameters, include parent resource selection
368
+ 4. Use ACTUAL URLs and paths from the codebase
369
+ 5. Write plain English steps - no technical jargon
370
+ 6. Base tests on REAL navigation flow, not assumptions
371
+
372
+ Now generate test specifications for the ticket above.`;
373
+ }
374
+ /**
375
+ * Agent Strategy Handling:
376
+ * - Claude: Uses tool calling with schema → guaranteed structured output
377
+ * - Cursor: Post-processes raw output with LangChain withStructuredOutput
378
+ *
379
+ * Node receives validated structured data regardless of agent type.
380
+ */
381
+