@illuma-ai/agents 1.1.21 → 1.1.22

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 (241) hide show
  1. package/dist/cjs/graphs/Graph.cjs +12 -1
  2. package/dist/cjs/graphs/Graph.cjs.map +1 -1
  3. package/dist/cjs/graphs/MultiAgentGraph.cjs +85 -1
  4. package/dist/cjs/graphs/MultiAgentGraph.cjs.map +1 -1
  5. package/dist/cjs/run.cjs +20 -9
  6. package/dist/cjs/run.cjs.map +1 -1
  7. package/dist/esm/graphs/Graph.mjs +12 -1
  8. package/dist/esm/graphs/Graph.mjs.map +1 -1
  9. package/dist/esm/graphs/MultiAgentGraph.mjs +85 -1
  10. package/dist/esm/graphs/MultiAgentGraph.mjs.map +1 -1
  11. package/dist/esm/run.mjs +20 -9
  12. package/dist/esm/run.mjs.map +1 -1
  13. package/dist/types/graphs/MultiAgentGraph.d.ts +17 -0
  14. package/package.json +1 -1
  15. package/src/graphs/Graph.ts +12 -1
  16. package/src/graphs/MultiAgentGraph.ts +105 -1
  17. package/src/graphs/__tests__/multi-agent-delegate.test.ts +191 -0
  18. package/src/run.ts +20 -11
  19. package/src/scripts/test-bedrock-handoff-autonomous.ts +231 -0
  20. package/src/agents/AgentContext.js +0 -782
  21. package/src/agents/AgentContext.test.js +0 -421
  22. package/src/agents/__tests__/AgentContext.test.js +0 -678
  23. package/src/agents/__tests__/resolveStructuredOutputMode.test.js +0 -117
  24. package/src/common/enum.js +0 -192
  25. package/src/common/index.js +0 -3
  26. package/src/events.js +0 -166
  27. package/src/graphs/Graph.js +0 -1857
  28. package/src/graphs/MultiAgentGraph.js +0 -1092
  29. package/src/graphs/__tests__/structured-output.integration.test.js +0 -624
  30. package/src/graphs/__tests__/structured-output.test.js +0 -144
  31. package/src/graphs/contextManagement.e2e.test.js +0 -718
  32. package/src/graphs/contextManagement.test.js +0 -485
  33. package/src/graphs/handoffValidation.test.js +0 -276
  34. package/src/graphs/index.js +0 -3
  35. package/src/index.js +0 -28
  36. package/src/instrumentation.js +0 -21
  37. package/src/llm/anthropic/index.js +0 -319
  38. package/src/llm/anthropic/types.js +0 -46
  39. package/src/llm/anthropic/utils/message_inputs.js +0 -627
  40. package/src/llm/anthropic/utils/message_outputs.js +0 -290
  41. package/src/llm/anthropic/utils/output_parsers.js +0 -89
  42. package/src/llm/anthropic/utils/tools.js +0 -25
  43. package/src/llm/bedrock/__tests__/bedrock-caching.test.js +0 -392
  44. package/src/llm/bedrock/index.js +0 -303
  45. package/src/llm/bedrock/types.js +0 -2
  46. package/src/llm/bedrock/utils/index.js +0 -6
  47. package/src/llm/bedrock/utils/message_inputs.js +0 -463
  48. package/src/llm/bedrock/utils/message_outputs.js +0 -269
  49. package/src/llm/fake.js +0 -92
  50. package/src/llm/google/index.js +0 -215
  51. package/src/llm/google/types.js +0 -12
  52. package/src/llm/google/utils/common.js +0 -670
  53. package/src/llm/google/utils/tools.js +0 -111
  54. package/src/llm/google/utils/zod_to_genai_parameters.js +0 -47
  55. package/src/llm/openai/index.js +0 -1033
  56. package/src/llm/openai/types.js +0 -2
  57. package/src/llm/openai/utils/index.js +0 -756
  58. package/src/llm/openai/utils/isReasoningModel.test.js +0 -79
  59. package/src/llm/openrouter/index.js +0 -261
  60. package/src/llm/openrouter/reasoning.test.js +0 -181
  61. package/src/llm/providers.js +0 -36
  62. package/src/llm/text.js +0 -65
  63. package/src/llm/vertexai/index.js +0 -402
  64. package/src/messages/__tests__/tools.test.js +0 -392
  65. package/src/messages/cache.js +0 -404
  66. package/src/messages/cache.test.js +0 -1167
  67. package/src/messages/content.js +0 -48
  68. package/src/messages/content.test.js +0 -314
  69. package/src/messages/core.js +0 -359
  70. package/src/messages/ensureThinkingBlock.test.js +0 -997
  71. package/src/messages/format.js +0 -973
  72. package/src/messages/formatAgentMessages.test.js +0 -2278
  73. package/src/messages/formatAgentMessages.tools.test.js +0 -362
  74. package/src/messages/formatMessage.test.js +0 -608
  75. package/src/messages/ids.js +0 -18
  76. package/src/messages/index.js +0 -9
  77. package/src/messages/labelContentByAgent.test.js +0 -725
  78. package/src/messages/prune.js +0 -438
  79. package/src/messages/reducer.js +0 -60
  80. package/src/messages/shiftIndexTokenCountMap.test.js +0 -63
  81. package/src/messages/summarize.js +0 -146
  82. package/src/messages/summarize.test.js +0 -332
  83. package/src/messages/tools.js +0 -90
  84. package/src/mockStream.js +0 -81
  85. package/src/prompts/collab.js +0 -7
  86. package/src/prompts/index.js +0 -3
  87. package/src/prompts/taskmanager.js +0 -58
  88. package/src/run.js +0 -427
  89. package/src/schemas/index.js +0 -3
  90. package/src/schemas/schema-preparation.test.js +0 -370
  91. package/src/schemas/validate.js +0 -314
  92. package/src/schemas/validate.test.js +0 -264
  93. package/src/scripts/abort.js +0 -127
  94. package/src/scripts/ant_web_search.js +0 -130
  95. package/src/scripts/ant_web_search_edge_case.js +0 -133
  96. package/src/scripts/ant_web_search_error_edge_case.js +0 -119
  97. package/src/scripts/args.js +0 -41
  98. package/src/scripts/bedrock-cache-debug.js +0 -186
  99. package/src/scripts/bedrock-content-aggregation-test.js +0 -195
  100. package/src/scripts/bedrock-merge-test.js +0 -80
  101. package/src/scripts/bedrock-parallel-tools-test.js +0 -150
  102. package/src/scripts/caching.js +0 -106
  103. package/src/scripts/cli.js +0 -152
  104. package/src/scripts/cli2.js +0 -119
  105. package/src/scripts/cli3.js +0 -163
  106. package/src/scripts/cli4.js +0 -165
  107. package/src/scripts/cli5.js +0 -165
  108. package/src/scripts/code_exec.js +0 -171
  109. package/src/scripts/code_exec_files.js +0 -180
  110. package/src/scripts/code_exec_multi_session.js +0 -185
  111. package/src/scripts/code_exec_ptc.js +0 -265
  112. package/src/scripts/code_exec_session.js +0 -217
  113. package/src/scripts/code_exec_simple.js +0 -120
  114. package/src/scripts/content.js +0 -111
  115. package/src/scripts/empty_input.js +0 -125
  116. package/src/scripts/handoff-test.js +0 -96
  117. package/src/scripts/image.js +0 -138
  118. package/src/scripts/memory.js +0 -83
  119. package/src/scripts/multi-agent-chain.js +0 -271
  120. package/src/scripts/multi-agent-conditional.js +0 -185
  121. package/src/scripts/multi-agent-document-review-chain.js +0 -171
  122. package/src/scripts/multi-agent-hybrid-flow.js +0 -264
  123. package/src/scripts/multi-agent-parallel-start.js +0 -214
  124. package/src/scripts/multi-agent-parallel.js +0 -346
  125. package/src/scripts/multi-agent-sequence.js +0 -184
  126. package/src/scripts/multi-agent-supervisor.js +0 -324
  127. package/src/scripts/multi-agent-test.js +0 -147
  128. package/src/scripts/parallel-asymmetric-tools-test.js +0 -202
  129. package/src/scripts/parallel-full-metadata-test.js +0 -176
  130. package/src/scripts/parallel-tools-test.js +0 -256
  131. package/src/scripts/programmatic_exec.js +0 -277
  132. package/src/scripts/programmatic_exec_agent.js +0 -168
  133. package/src/scripts/search.js +0 -118
  134. package/src/scripts/sequential-full-metadata-test.js +0 -143
  135. package/src/scripts/simple.js +0 -174
  136. package/src/scripts/single-agent-metadata-test.js +0 -152
  137. package/src/scripts/stream.js +0 -113
  138. package/src/scripts/test-custom-prompt-key.js +0 -132
  139. package/src/scripts/test-handoff-input.js +0 -143
  140. package/src/scripts/test-handoff-preamble.js +0 -227
  141. package/src/scripts/test-handoff-steering.js +0 -353
  142. package/src/scripts/test-multi-agent-list-handoff.js +0 -318
  143. package/src/scripts/test-parallel-agent-labeling.js +0 -253
  144. package/src/scripts/test-parallel-handoffs.js +0 -229
  145. package/src/scripts/test-thinking-handoff-bedrock.js +0 -132
  146. package/src/scripts/test-thinking-handoff.js +0 -132
  147. package/src/scripts/test-thinking-to-thinking-handoff-bedrock.js +0 -140
  148. package/src/scripts/test-tool-before-handoff-role-order.js +0 -223
  149. package/src/scripts/test-tools-before-handoff.js +0 -187
  150. package/src/scripts/test_code_api.js +0 -263
  151. package/src/scripts/thinking-bedrock.js +0 -128
  152. package/src/scripts/thinking-vertexai.js +0 -130
  153. package/src/scripts/thinking.js +0 -134
  154. package/src/scripts/tool_search.js +0 -114
  155. package/src/scripts/tools.js +0 -125
  156. package/src/specs/agent-handoffs-bedrock.integration.test.js +0 -280
  157. package/src/specs/agent-handoffs.test.js +0 -924
  158. package/src/specs/anthropic.simple.test.js +0 -287
  159. package/src/specs/azure.simple.test.js +0 -381
  160. package/src/specs/cache.simple.test.js +0 -282
  161. package/src/specs/custom-event-await.test.js +0 -148
  162. package/src/specs/deepseek.simple.test.js +0 -189
  163. package/src/specs/emergency-prune.test.js +0 -308
  164. package/src/specs/moonshot.simple.test.js +0 -237
  165. package/src/specs/observability.integration.test.js +0 -1337
  166. package/src/specs/openai.simple.test.js +0 -233
  167. package/src/specs/openrouter.simple.test.js +0 -202
  168. package/src/specs/prune.test.js +0 -733
  169. package/src/specs/reasoning.test.js +0 -144
  170. package/src/specs/spec.utils.js +0 -4
  171. package/src/specs/thinking-handoff.test.js +0 -486
  172. package/src/specs/thinking-prune.test.js +0 -600
  173. package/src/specs/token-distribution-edge-case.test.js +0 -246
  174. package/src/specs/token-memoization.test.js +0 -32
  175. package/src/specs/tokens.test.js +0 -49
  176. package/src/specs/tool-error.test.js +0 -139
  177. package/src/splitStream.js +0 -204
  178. package/src/splitStream.test.js +0 -504
  179. package/src/stream.js +0 -650
  180. package/src/stream.test.js +0 -225
  181. package/src/test/mockTools.js +0 -340
  182. package/src/tools/BrowserTools.js +0 -245
  183. package/src/tools/Calculator.js +0 -38
  184. package/src/tools/Calculator.test.js +0 -225
  185. package/src/tools/CodeExecutor.js +0 -233
  186. package/src/tools/ProgrammaticToolCalling.js +0 -602
  187. package/src/tools/StreamingToolCallBuffer.js +0 -179
  188. package/src/tools/ToolNode.js +0 -930
  189. package/src/tools/ToolSearch.js +0 -904
  190. package/src/tools/__tests__/BrowserTools.test.js +0 -306
  191. package/src/tools/__tests__/ProgrammaticToolCalling.integration.test.js +0 -276
  192. package/src/tools/__tests__/ProgrammaticToolCalling.test.js +0 -807
  193. package/src/tools/__tests__/StreamingToolCallBuffer.test.js +0 -175
  194. package/src/tools/__tests__/ToolApproval.test.js +0 -675
  195. package/src/tools/__tests__/ToolNode.recovery.test.js +0 -200
  196. package/src/tools/__tests__/ToolNode.session.test.js +0 -319
  197. package/src/tools/__tests__/ToolSearch.integration.test.js +0 -125
  198. package/src/tools/__tests__/ToolSearch.test.js +0 -812
  199. package/src/tools/__tests__/handlers.test.js +0 -799
  200. package/src/tools/__tests__/truncation-recovery.integration.test.js +0 -362
  201. package/src/tools/handlers.js +0 -306
  202. package/src/tools/schema.js +0 -25
  203. package/src/tools/search/anthropic.js +0 -34
  204. package/src/tools/search/content.js +0 -116
  205. package/src/tools/search/content.test.js +0 -133
  206. package/src/tools/search/firecrawl.js +0 -173
  207. package/src/tools/search/format.js +0 -198
  208. package/src/tools/search/highlights.js +0 -241
  209. package/src/tools/search/index.js +0 -3
  210. package/src/tools/search/jina-reranker.test.js +0 -106
  211. package/src/tools/search/rerankers.js +0 -165
  212. package/src/tools/search/schema.js +0 -102
  213. package/src/tools/search/search.js +0 -561
  214. package/src/tools/search/serper-scraper.js +0 -126
  215. package/src/tools/search/test.js +0 -129
  216. package/src/tools/search/tool.js +0 -453
  217. package/src/tools/search/types.js +0 -2
  218. package/src/tools/search/utils.js +0 -59
  219. package/src/types/graph.js +0 -24
  220. package/src/types/graph.test.js +0 -192
  221. package/src/types/index.js +0 -7
  222. package/src/types/llm.js +0 -2
  223. package/src/types/messages.js +0 -2
  224. package/src/types/run.js +0 -2
  225. package/src/types/stream.js +0 -2
  226. package/src/types/tools.js +0 -2
  227. package/src/utils/contextAnalytics.js +0 -79
  228. package/src/utils/contextAnalytics.test.js +0 -166
  229. package/src/utils/events.js +0 -26
  230. package/src/utils/graph.js +0 -11
  231. package/src/utils/handlers.js +0 -65
  232. package/src/utils/index.js +0 -10
  233. package/src/utils/llm.js +0 -21
  234. package/src/utils/llmConfig.js +0 -205
  235. package/src/utils/logging.js +0 -37
  236. package/src/utils/misc.js +0 -51
  237. package/src/utils/run.js +0 -69
  238. package/src/utils/schema.js +0 -21
  239. package/src/utils/title.js +0 -119
  240. package/src/utils/tokens.js +0 -92
  241. package/src/utils/toonFormat.js +0 -379
@@ -1,200 +0,0 @@
1
- import { z } from 'zod';
2
- import { tool } from '@langchain/core/tools';
3
- import { AIMessage } from '@langchain/core/messages';
4
- import { describe, it, expect } from '@jest/globals';
5
- import { ToolNode } from '../ToolNode';
6
- import { StreamingToolCallBuffer } from '../StreamingToolCallBuffer';
7
- /**
8
- * Creates a mock content_tool that captures the args it receives.
9
- * Used to verify that truncated args are recovered from the streaming buffer.
10
- */
11
- function createMockContentTool(capturedArgs) {
12
- return tool(async (input) => {
13
- capturedArgs.push({ ...input });
14
- return `Wrote ${input.filename ?? 'file'}`;
15
- }, {
16
- name: 'content_tool',
17
- description: 'Read, write, and manage files',
18
- schema: z.object({
19
- action: z.string(),
20
- filename: z.string().optional(),
21
- content: z.string().optional(),
22
- }),
23
- });
24
- }
25
- function createAIMessageWithContentTool(callId, args) {
26
- return new AIMessage({
27
- content: '',
28
- tool_calls: [
29
- {
30
- id: callId,
31
- name: 'content_tool',
32
- args,
33
- },
34
- ],
35
- });
36
- }
37
- describe('ToolNode truncated arg recovery', () => {
38
- it('recovers missing content field from streaming buffer', async () => {
39
- const capturedArgs = [];
40
- const buffer = new StreamingToolCallBuffer();
41
- // Simulate streaming: buffer accumulated the full raw args
42
- buffer.append('tc_1', '{"action":"write","filename":"app.tsx","content":"function App() {\\n return <div>Hello</div>;\\n}"}');
43
- buffer.setToolName('tc_1', 'content_tool');
44
- const contentTool = createMockContentTool(capturedArgs);
45
- const toolNode = new ToolNode({
46
- tools: [contentTool],
47
- streamingToolCallBuffer: buffer,
48
- });
49
- // Simulate truncated args: parsePartialJson lost the content field
50
- const aiMsg = createAIMessageWithContentTool('tc_1', {
51
- action: 'write',
52
- filename: 'app.tsx',
53
- // content is MISSING — truncated by max_tokens
54
- });
55
- await toolNode.invoke({ messages: [aiMsg] });
56
- expect(capturedArgs).toHaveLength(1);
57
- expect(capturedArgs[0].action).toBe('write');
58
- expect(capturedArgs[0].filename).toBe('app.tsx');
59
- expect(capturedArgs[0].content).toBe('function App() {\n return <div>Hello</div>;\n}');
60
- });
61
- it('recovers completely empty args from streaming buffer', async () => {
62
- const capturedArgs = [];
63
- const buffer = new StreamingToolCallBuffer();
64
- // Buffer has full args, but parsePartialJson returned empty object
65
- buffer.append('tc_1', '{"action":"write","filename":"test.js","content":"const x = 1;"}');
66
- buffer.setToolName('tc_1', 'content_tool');
67
- const contentTool = createMockContentTool(capturedArgs);
68
- const toolNode = new ToolNode({
69
- tools: [contentTool],
70
- streamingToolCallBuffer: buffer,
71
- });
72
- // Empty args — JSON parsing aborted completely
73
- const aiMsg = createAIMessageWithContentTool('tc_1', {});
74
- await toolNode.invoke({ messages: [aiMsg] });
75
- expect(capturedArgs).toHaveLength(1);
76
- expect(capturedArgs[0].action).toBe('write');
77
- expect(capturedArgs[0].filename).toBe('test.js');
78
- expect(capturedArgs[0].content).toBe('const x = 1;');
79
- });
80
- it('does not modify args when buffer has no entry', async () => {
81
- const capturedArgs = [];
82
- const buffer = new StreamingToolCallBuffer();
83
- // No buffer entry for tc_1
84
- const contentTool = createMockContentTool(capturedArgs);
85
- const toolNode = new ToolNode({
86
- tools: [contentTool],
87
- streamingToolCallBuffer: buffer,
88
- });
89
- const aiMsg = createAIMessageWithContentTool('tc_1', {
90
- action: 'read',
91
- filename: 'test.js',
92
- });
93
- await toolNode.invoke({ messages: [aiMsg] });
94
- expect(capturedArgs).toHaveLength(1);
95
- expect(capturedArgs[0].action).toBe('read');
96
- expect(capturedArgs[0].filename).toBe('test.js');
97
- expect(capturedArgs[0].content).toBeUndefined();
98
- });
99
- it('does not overwrite existing parsed values', async () => {
100
- const capturedArgs = [];
101
- const buffer = new StreamingToolCallBuffer();
102
- // Buffer has different values than parsed
103
- buffer.append('tc_1', '{"action":"write","filename":"old.tsx","content":"old content"}');
104
- const contentTool = createMockContentTool(capturedArgs);
105
- const toolNode = new ToolNode({
106
- tools: [contentTool],
107
- streamingToolCallBuffer: buffer,
108
- });
109
- // Parsed args have action and filename already
110
- const aiMsg = createAIMessageWithContentTool('tc_1', {
111
- action: 'write',
112
- filename: 'new.tsx',
113
- // content is missing — should be recovered
114
- });
115
- await toolNode.invoke({ messages: [aiMsg] });
116
- expect(capturedArgs).toHaveLength(1);
117
- expect(capturedArgs[0].action).toBe('write');
118
- expect(capturedArgs[0].filename).toBe('new.tsx'); // Keeps parsed value, not buffer value
119
- expect(capturedArgs[0].content).toBe('old content'); // Recovered from buffer
120
- });
121
- it('clears buffer after recovery', async () => {
122
- const capturedArgs = [];
123
- const buffer = new StreamingToolCallBuffer();
124
- buffer.append('tc_1', '{"action":"write","content":"test"}');
125
- const contentTool = createMockContentTool(capturedArgs);
126
- const toolNode = new ToolNode({
127
- tools: [contentTool],
128
- streamingToolCallBuffer: buffer,
129
- });
130
- const aiMsg = createAIMessageWithContentTool('tc_1', { action: 'write' });
131
- await toolNode.invoke({ messages: [aiMsg] });
132
- // Buffer should be cleared after processing
133
- expect(buffer.has('tc_1')).toBe(false);
134
- });
135
- it('works with tools other than content_tool', async () => {
136
- const capturedArgs = [];
137
- const buffer = new StreamingToolCallBuffer();
138
- buffer.append('tc_1', '{"query":"SELECT * FROM users","database":"prod"}');
139
- buffer.setToolName('tc_1', 'sql_query');
140
- const sqlTool = tool(async (input) => {
141
- capturedArgs.push({ ...input });
142
- return 'Results: 42 rows';
143
- }, {
144
- name: 'sql_query',
145
- description: 'Execute SQL query',
146
- schema: z.object({
147
- query: z.string(),
148
- database: z.string().optional(),
149
- }),
150
- });
151
- const toolNode = new ToolNode({
152
- tools: [sqlTool],
153
- streamingToolCallBuffer: buffer,
154
- });
155
- // Truncated: only has query, missing database
156
- const aiMsg = new AIMessage({
157
- content: '',
158
- tool_calls: [
159
- { id: 'tc_1', name: 'sql_query', args: { query: 'SELECT * FROM users' } },
160
- ],
161
- });
162
- await toolNode.invoke({ messages: [aiMsg] });
163
- expect(capturedArgs).toHaveLength(1);
164
- expect(capturedArgs[0].query).toBe('SELECT * FROM users');
165
- expect(capturedArgs[0].database).toBe('prod');
166
- });
167
- it('recovers truncated content with newlines and special chars', async () => {
168
- const capturedArgs = [];
169
- const buffer = new StreamingToolCallBuffer();
170
- const largeContent = 'import React from \\"react\\";\\n\\n' +
171
- 'export default function Dashboard() {\\n' +
172
- ' const [count, setCount] = React.useState(0);\\n' +
173
- ' return (\\n' +
174
- ' <div className=\\"dashboard\\">\\n' +
175
- ' <h1>Dashboard</h1>\\n' +
176
- ' <button onClick={() => setCount(c => c + 1)}>\\n' +
177
- ' Count: {count}\\n' +
178
- ' </button>\\n' +
179
- ' </div>\\n' +
180
- ' );\\n' +
181
- '}';
182
- buffer.append('tc_1', `{"action":"write","filename":"Dashboard.tsx","content":"${largeContent}"}`);
183
- const contentTool = createMockContentTool(capturedArgs);
184
- const toolNode = new ToolNode({
185
- tools: [contentTool],
186
- streamingToolCallBuffer: buffer,
187
- });
188
- // content_tool with missing content
189
- const aiMsg = createAIMessageWithContentTool('tc_1', {
190
- action: 'write',
191
- filename: 'Dashboard.tsx',
192
- });
193
- await toolNode.invoke({ messages: [aiMsg] });
194
- expect(capturedArgs).toHaveLength(1);
195
- expect(capturedArgs[0].content).toContain('import React from "react"');
196
- expect(capturedArgs[0].content).toContain('export default function Dashboard()');
197
- expect(capturedArgs[0].content).toContain('Count: {count}');
198
- });
199
- });
200
- //# sourceMappingURL=ToolNode.recovery.test.js.map
@@ -1,319 +0,0 @@
1
- import { z } from 'zod';
2
- import { tool } from '@langchain/core/tools';
3
- import { AIMessage } from '@langchain/core/messages';
4
- import { describe, it, expect } from '@jest/globals';
5
- import { ToolNode } from '../ToolNode';
6
- import { Constants } from '@/common';
7
- /**
8
- * Creates a mock execute_code tool that captures the toolCall config it receives.
9
- * Returns a content_and_artifact response with configurable session/files.
10
- */
11
- function createMockCodeTool(options) {
12
- const { capturedConfigs, artifact } = options;
13
- const defaultArtifact = {
14
- session_id: 'new-session-123',
15
- files: [],
16
- };
17
- return tool(async (_input, config) => {
18
- capturedConfigs.push({ ...(config.toolCall ?? {}) });
19
- return ['stdout:\nhello world\n', artifact ?? defaultArtifact];
20
- }, {
21
- name: Constants.EXECUTE_CODE,
22
- description: 'Execute code in a sandbox',
23
- schema: z.object({ lang: z.string(), code: z.string() }),
24
- responseFormat: Constants.CONTENT_AND_ARTIFACT,
25
- });
26
- }
27
- function createAIMessageWithCodeCall(callId) {
28
- return new AIMessage({
29
- content: '',
30
- tool_calls: [
31
- {
32
- id: callId,
33
- name: Constants.EXECUTE_CODE,
34
- args: { lang: 'python', code: 'print("hello")' },
35
- },
36
- ],
37
- });
38
- }
39
- describe('ToolNode code execution session management', () => {
40
- describe('session injection via runTool (direct execution)', () => {
41
- it('injects session_id and _injected_files when session has files', async () => {
42
- const capturedConfigs = [];
43
- const sessions = new Map();
44
- sessions.set(Constants.EXECUTE_CODE, {
45
- session_id: 'prev-session-abc',
46
- files: [
47
- { id: 'file1', name: 'data.csv', session_id: 'prev-session-abc' },
48
- { id: 'file2', name: 'chart.png', session_id: 'prev-session-abc' },
49
- ],
50
- lastUpdated: Date.now(),
51
- });
52
- const mockTool = createMockCodeTool({ capturedConfigs });
53
- const toolNode = new ToolNode({ tools: [mockTool], sessions });
54
- const aiMsg = createAIMessageWithCodeCall('call_1');
55
- await toolNode.invoke({ messages: [aiMsg] });
56
- expect(capturedConfigs).toHaveLength(1);
57
- expect(capturedConfigs[0].session_id).toBe('prev-session-abc');
58
- expect(capturedConfigs[0]._injected_files).toEqual([
59
- { session_id: 'prev-session-abc', id: 'file1', name: 'data.csv' },
60
- { session_id: 'prev-session-abc', id: 'file2', name: 'chart.png' },
61
- ]);
62
- });
63
- it('injects session_id even when session has no tracked files', async () => {
64
- const capturedConfigs = [];
65
- const sessions = new Map();
66
- sessions.set(Constants.EXECUTE_CODE, {
67
- session_id: 'prev-session-no-files',
68
- files: [],
69
- lastUpdated: Date.now(),
70
- });
71
- const mockTool = createMockCodeTool({ capturedConfigs });
72
- const toolNode = new ToolNode({ tools: [mockTool], sessions });
73
- const aiMsg = createAIMessageWithCodeCall('call_2');
74
- await toolNode.invoke({ messages: [aiMsg] });
75
- expect(capturedConfigs).toHaveLength(1);
76
- expect(capturedConfigs[0].session_id).toBe('prev-session-no-files');
77
- expect(capturedConfigs[0]._injected_files).toBeUndefined();
78
- });
79
- it('does not inject session context when no session exists', async () => {
80
- const capturedConfigs = [];
81
- const sessions = new Map();
82
- const mockTool = createMockCodeTool({ capturedConfigs });
83
- const toolNode = new ToolNode({ tools: [mockTool], sessions });
84
- const aiMsg = createAIMessageWithCodeCall('call_3');
85
- await toolNode.invoke({ messages: [aiMsg] });
86
- expect(capturedConfigs).toHaveLength(1);
87
- expect(capturedConfigs[0].session_id).toBeUndefined();
88
- expect(capturedConfigs[0]._injected_files).toBeUndefined();
89
- });
90
- it('preserves per-file session_id for multi-session files', async () => {
91
- const capturedConfigs = [];
92
- const sessions = new Map();
93
- sessions.set(Constants.EXECUTE_CODE, {
94
- session_id: 'session-B',
95
- files: [
96
- { id: 'f1', name: 'old.csv', session_id: 'session-A' },
97
- { id: 'f2', name: 'new.png', session_id: 'session-B' },
98
- ],
99
- lastUpdated: Date.now(),
100
- });
101
- const mockTool = createMockCodeTool({ capturedConfigs });
102
- const toolNode = new ToolNode({ tools: [mockTool], sessions });
103
- const aiMsg = createAIMessageWithCodeCall('call_4');
104
- await toolNode.invoke({ messages: [aiMsg] });
105
- const files = capturedConfigs[0]._injected_files;
106
- expect(files[0].session_id).toBe('session-A');
107
- expect(files[1].session_id).toBe('session-B');
108
- });
109
- });
110
- describe('getCodeSessionContext (via dispatchToolEvents request building)', () => {
111
- it('builds session context with files for event-driven requests', () => {
112
- const sessions = new Map();
113
- sessions.set(Constants.EXECUTE_CODE, {
114
- session_id: 'evt-session',
115
- files: [{ id: 'ef1', name: 'out.parquet', session_id: 'evt-session' }],
116
- lastUpdated: Date.now(),
117
- });
118
- const mockTool = createMockCodeTool({ capturedConfigs: [] });
119
- const toolNode = new ToolNode({
120
- tools: [mockTool],
121
- sessions,
122
- eventDrivenMode: true,
123
- });
124
- const context = toolNode.getCodeSessionContext();
125
- expect(context).toEqual({
126
- session_id: 'evt-session',
127
- files: [{ session_id: 'evt-session', id: 'ef1', name: 'out.parquet' }],
128
- });
129
- });
130
- it('builds session context without files when session has no tracked files', () => {
131
- const sessions = new Map();
132
- sessions.set(Constants.EXECUTE_CODE, {
133
- session_id: 'evt-session-empty',
134
- files: [],
135
- lastUpdated: Date.now(),
136
- });
137
- const mockTool = createMockCodeTool({ capturedConfigs: [] });
138
- const toolNode = new ToolNode({
139
- tools: [mockTool],
140
- sessions,
141
- eventDrivenMode: true,
142
- });
143
- const context = toolNode.getCodeSessionContext();
144
- expect(context).toEqual({ session_id: 'evt-session-empty' });
145
- });
146
- it('returns undefined when no session exists', () => {
147
- const sessions = new Map();
148
- const mockTool = createMockCodeTool({ capturedConfigs: [] });
149
- const toolNode = new ToolNode({
150
- tools: [mockTool],
151
- sessions,
152
- eventDrivenMode: true,
153
- });
154
- const context = toolNode.getCodeSessionContext();
155
- expect(context).toBeUndefined();
156
- });
157
- });
158
- describe('storeCodeSessionFromResults (session storage from artifacts)', () => {
159
- it('stores session with files from code execution results', () => {
160
- const sessions = new Map();
161
- const mockTool = createMockCodeTool({ capturedConfigs: [] });
162
- const toolNode = new ToolNode({
163
- tools: [mockTool],
164
- sessions,
165
- eventDrivenMode: true,
166
- });
167
- const storeMethod = toolNode.storeCodeSessionFromResults.bind(toolNode);
168
- storeMethod([
169
- {
170
- toolCallId: 'tc1',
171
- content: 'output',
172
- artifact: {
173
- session_id: 'new-sess',
174
- files: [{ id: 'f1', name: 'result.csv' }],
175
- },
176
- status: 'success',
177
- },
178
- ], [{ id: 'tc1', name: Constants.EXECUTE_CODE, args: {} }]);
179
- const stored = sessions.get(Constants.EXECUTE_CODE);
180
- expect(stored).toBeDefined();
181
- expect(stored.session_id).toBe('new-sess');
182
- expect(stored.files).toHaveLength(1);
183
- expect(stored.files[0]).toEqual(expect.objectContaining({
184
- id: 'f1',
185
- name: 'result.csv',
186
- session_id: 'new-sess',
187
- }));
188
- });
189
- it('stores session_id even when Code API returns no files', () => {
190
- const sessions = new Map();
191
- const mockTool = createMockCodeTool({ capturedConfigs: [] });
192
- const toolNode = new ToolNode({
193
- tools: [mockTool],
194
- sessions,
195
- eventDrivenMode: true,
196
- });
197
- const storeMethod = toolNode.storeCodeSessionFromResults.bind(toolNode);
198
- storeMethod([
199
- {
200
- toolCallId: 'tc2',
201
- content: 'stdout:\nSaved parquet\n',
202
- artifact: { session_id: 'parquet-session', files: [] },
203
- status: 'success',
204
- },
205
- ], [{ id: 'tc2', name: Constants.EXECUTE_CODE, args: {} }]);
206
- const stored = sessions.get(Constants.EXECUTE_CODE);
207
- expect(stored).toBeDefined();
208
- expect(stored.session_id).toBe('parquet-session');
209
- expect(stored.files).toEqual([]);
210
- });
211
- it('merges new files with existing session, replacing same-name files', () => {
212
- const sessions = new Map();
213
- sessions.set(Constants.EXECUTE_CODE, {
214
- session_id: 'old-sess',
215
- files: [
216
- { id: 'f1', name: 'data.csv', session_id: 'old-sess' },
217
- { id: 'f2', name: 'chart.png', session_id: 'old-sess' },
218
- ],
219
- lastUpdated: Date.now(),
220
- });
221
- const mockTool = createMockCodeTool({ capturedConfigs: [] });
222
- const toolNode = new ToolNode({
223
- tools: [mockTool],
224
- sessions,
225
- eventDrivenMode: true,
226
- });
227
- const storeMethod = toolNode.storeCodeSessionFromResults.bind(toolNode);
228
- storeMethod([
229
- {
230
- toolCallId: 'tc3',
231
- content: 'output',
232
- artifact: {
233
- session_id: 'new-sess',
234
- files: [{ id: 'f3', name: 'chart.png' }],
235
- },
236
- status: 'success',
237
- },
238
- ], [{ id: 'tc3', name: Constants.EXECUTE_CODE, args: {} }]);
239
- const stored = sessions.get(Constants.EXECUTE_CODE);
240
- expect(stored.session_id).toBe('new-sess');
241
- expect(stored.files).toHaveLength(2);
242
- const csvFile = stored.files.find((f) => f.name === 'data.csv');
243
- expect(csvFile.session_id).toBe('old-sess');
244
- const chartFile = stored.files.find((f) => f.name === 'chart.png');
245
- expect(chartFile.id).toBe('f3');
246
- expect(chartFile.session_id).toBe('new-sess');
247
- });
248
- it('preserves existing files when new execution has no files', () => {
249
- const sessions = new Map();
250
- sessions.set(Constants.EXECUTE_CODE, {
251
- session_id: 'old-sess',
252
- files: [{ id: 'f1', name: 'data.csv', session_id: 'old-sess' }],
253
- lastUpdated: Date.now(),
254
- });
255
- const mockTool = createMockCodeTool({ capturedConfigs: [] });
256
- const toolNode = new ToolNode({
257
- tools: [mockTool],
258
- sessions,
259
- eventDrivenMode: true,
260
- });
261
- const storeMethod = toolNode.storeCodeSessionFromResults.bind(toolNode);
262
- storeMethod([
263
- {
264
- toolCallId: 'tc4',
265
- content: 'stdout:\nno files generated\n',
266
- artifact: { session_id: 'new-sess', files: [] },
267
- status: 'success',
268
- },
269
- ], [{ id: 'tc4', name: Constants.EXECUTE_CODE, args: {} }]);
270
- const stored = sessions.get(Constants.EXECUTE_CODE);
271
- expect(stored.session_id).toBe('new-sess');
272
- expect(stored.files).toHaveLength(1);
273
- expect(stored.files[0].name).toBe('data.csv');
274
- });
275
- it('ignores non-code-execution tool results', () => {
276
- const sessions = new Map();
277
- const mockTool = createMockCodeTool({ capturedConfigs: [] });
278
- const toolNode = new ToolNode({
279
- tools: [mockTool],
280
- sessions,
281
- eventDrivenMode: true,
282
- });
283
- const storeMethod = toolNode.storeCodeSessionFromResults.bind(toolNode);
284
- storeMethod([
285
- {
286
- toolCallId: 'tc5',
287
- content: 'search results',
288
- artifact: { session_id: 'should-not-store' },
289
- status: 'success',
290
- },
291
- ], [{ id: 'tc5', name: 'web_search', args: {} }]);
292
- expect(sessions.has(Constants.EXECUTE_CODE)).toBe(false);
293
- });
294
- it('ignores error results', () => {
295
- const sessions = new Map();
296
- const mockTool = createMockCodeTool({ capturedConfigs: [] });
297
- const toolNode = new ToolNode({
298
- tools: [mockTool],
299
- sessions,
300
- eventDrivenMode: true,
301
- });
302
- const storeMethod = toolNode.storeCodeSessionFromResults.bind(toolNode);
303
- storeMethod([
304
- {
305
- toolCallId: 'tc6',
306
- content: '',
307
- artifact: {
308
- session_id: 'error-session',
309
- files: [{ id: 'f1', name: 'x' }],
310
- },
311
- status: 'error',
312
- errorMessage: 'execution failed',
313
- },
314
- ], [{ id: 'tc6', name: Constants.EXECUTE_CODE, args: {} }]);
315
- expect(sessions.has(Constants.EXECUTE_CODE)).toBe(false);
316
- });
317
- });
318
- });
319
- //# sourceMappingURL=ToolNode.session.test.js.map
@@ -1,125 +0,0 @@
1
- // src/tools/__tests__/ToolSearch.integration.test.ts
2
- /**
3
- * Integration tests for Tool Search Regex.
4
- * These tests hit the LIVE Code API and verify end-to-end search functionality.
5
- *
6
- * Run with: npm test -- ToolSearch.integration.test.ts
7
- *
8
- * Requires CODE_EXECUTOR_API_KEY environment variable.
9
- * Tests are skipped when the API key is not available.
10
- */
11
- import { config as dotenvConfig } from 'dotenv';
12
- dotenvConfig();
13
- import { describe, it, expect, beforeAll } from '@jest/globals';
14
- import { createToolSearch } from '../ToolSearch';
15
- import { createToolSearchToolRegistry } from '@/test/mockTools';
16
- const apiKey = process.env.CODE_EXECUTOR_API_KEY;
17
- const shouldSkip = apiKey == null || apiKey === '';
18
- const describeIfApiKey = shouldSkip ? describe.skip : describe;
19
- describeIfApiKey('ToolSearch - Live API Integration', () => {
20
- let searchTool;
21
- const toolRegistry = createToolSearchToolRegistry();
22
- beforeAll(() => {
23
- searchTool = createToolSearch({ apiKey: apiKey, toolRegistry });
24
- });
25
- it('searches for expense-related tools', async () => {
26
- const output = await searchTool.invoke({ query: 'expense' });
27
- expect(typeof output).toBe('string');
28
- expect(output).toContain('Found');
29
- expect(output).toContain('matching tools');
30
- expect(output).toContain('get_expenses');
31
- expect(output).toContain('calculate_expense_totals');
32
- expect(output).toContain('score: 0.95');
33
- }, 10000);
34
- it('searches for weather tools with OR pattern', async () => {
35
- const output = await searchTool.invoke({ query: 'weather|forecast' });
36
- expect(output).toContain('Found');
37
- expect(output).toContain('get_weather');
38
- expect(output).toContain('get_forecast');
39
- }, 10000);
40
- it('performs case-insensitive search', async () => {
41
- const output = await searchTool.invoke({ query: 'EMAIL' });
42
- expect(output).toContain('send_email');
43
- }, 10000);
44
- it('searches descriptions only', async () => {
45
- const output = await searchTool.invoke({
46
- query: 'database',
47
- fields: ['description'],
48
- });
49
- expect(output).toContain('Found');
50
- expect(output).toContain('Matched in: description');
51
- expect(output).toContain('run_database_query');
52
- }, 10000);
53
- it('searches parameter names', async () => {
54
- const output = await searchTool.invoke({
55
- query: 'query',
56
- fields: ['parameters'],
57
- });
58
- expect(output).toContain('Found');
59
- expect(output).toContain('Matched in: parameters');
60
- }, 10000);
61
- it('limits results to specified count', async () => {
62
- const output = await searchTool.invoke({
63
- query: 'get',
64
- max_results: 2,
65
- });
66
- expect(output).toContain('Found 2 matching tools');
67
- const toolMatches = output.match(/- \w+ \(score:/g);
68
- expect(toolMatches?.length).toBe(2);
69
- }, 10000);
70
- it('returns no matches for nonsense pattern', async () => {
71
- const output = await searchTool.invoke({
72
- query: 'xyznonexistent999',
73
- });
74
- expect(output).toContain('No tools matched');
75
- expect(output).toContain('Total tools searched:');
76
- }, 10000);
77
- it('uses regex character classes', async () => {
78
- const output = await searchTool.invoke({ query: 'get_[a-z]+' });
79
- expect(output).toContain('Found');
80
- expect(output).toContain('matching tools');
81
- }, 10000);
82
- it('sanitizes dangerous patterns with warning', async () => {
83
- const output = await searchTool.invoke({ query: '(a+)+' });
84
- expect(output).toContain('Note: The provided pattern was converted to a literal search for safety');
85
- // After sanitization, pattern is shown in the output
86
- expect(output).toContain('Total tools searched:');
87
- expect(output).toContain('No tools matched');
88
- }, 10000);
89
- it('searches across all fields', async () => {
90
- const output = await searchTool.invoke({
91
- query: 'text',
92
- fields: ['name', 'description', 'parameters'],
93
- });
94
- expect(output).toContain('Found');
95
- expect(output).toContain('translate_text');
96
- }, 10000);
97
- it('handles complex regex patterns', async () => {
98
- const output = await searchTool.invoke({
99
- query: '^(get|create)_.*',
100
- });
101
- expect(output).toContain('Found');
102
- expect(output).toContain('matching tools');
103
- }, 10000);
104
- it('prioritizes name matches over description matches', async () => {
105
- const output = await searchTool.invoke({ query: 'generate' });
106
- expect(output).toContain('generate_report');
107
- expect(output).toContain('score:');
108
- const lines = output.split('\n');
109
- const generateLine = lines.find((l) => l.includes('generate_report'));
110
- expect(generateLine).toContain('score: 0.95'); // Name match = 0.95
111
- }, 10000);
112
- it('finds tools by partial name match', async () => {
113
- const output = await searchTool.invoke({ query: 'budget' });
114
- expect(output).toContain('create_budget');
115
- }, 10000);
116
- it('handles very specific patterns', async () => {
117
- const output = await searchTool.invoke({
118
- query: 'send_email',
119
- });
120
- expect(output).toContain('Found');
121
- expect(output).toContain('send_email');
122
- expect(output).toContain('score: 0.95');
123
- }, 10000);
124
- });
125
- //# sourceMappingURL=ToolSearch.integration.test.js.map