engineering-behavior-observatory 0.2.1

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 (178) hide show
  1. package/LICENSE +202 -0
  2. package/README.md +151 -0
  3. package/contracts/codex-app-server-0.150.1/manifest.json +25 -0
  4. package/contracts/codex-app-server-0.150.1/schema/InitializeParams.json +84 -0
  5. package/contracts/codex-app-server-0.150.1/schema/JSONRPCMessage.json +137 -0
  6. package/contracts/codex-app-server-0.150.1/schema/v2/ThreadReadParams.json +17 -0
  7. package/contracts/codex-app-server-0.150.1/schema/v2/ThreadTokenUsageUpdatedNotification.json +82 -0
  8. package/contracts/codex-app-server-0.150.1/schema/v2/TurnInterruptParams.json +17 -0
  9. package/contracts/codex-app-server-0.150.1/types/AskForApproval.ts +5 -0
  10. package/contracts/codex-app-server-0.150.1/types/ClientNotification.ts +5 -0
  11. package/contracts/codex-app-server-0.150.1/types/SandboxMode.ts +5 -0
  12. package/contracts/codex-app-server-0.150.1/types/ThreadReadParams.ts +9 -0
  13. package/contracts/codex-app-server-0.150.1/types/TokenUsageBreakdown.ts +5 -0
  14. package/contracts/codex-app-server-0.150.1/types/TurnInterruptParams.ts +5 -0
  15. package/contracts/codex-app-server-0.153.4/manifest.json +27 -0
  16. package/contracts/codex-app-server-0.153.4/schema/JSONRPCMessage.json +137 -0
  17. package/contracts/codex-app-server-0.153.4/schema/v1/InitializeParams.json +84 -0
  18. package/contracts/codex-app-server-0.153.4/schema/v2/ThreadReadParams.json +17 -0
  19. package/contracts/codex-app-server-0.153.4/schema/v2/ThreadStartParams.json +515 -0
  20. package/contracts/codex-app-server-0.153.4/schema/v2/ThreadTokenUsageUpdatedNotification.json +82 -0
  21. package/contracts/codex-app-server-0.153.4/schema/v2/TurnInterruptParams.json +17 -0
  22. package/contracts/codex-app-server-0.153.4/schema/v2/TurnStartParams.json +921 -0
  23. package/contracts/codex-app-server-0.153.4/types/AskForApproval.ts +5 -0
  24. package/contracts/codex-app-server-0.153.4/types/ClientNotification.ts +5 -0
  25. package/contracts/codex-app-server-0.153.4/types/SandboxMode.ts +5 -0
  26. package/contracts/codex-app-server-0.153.4/types/ThreadReadParams.ts +12 -0
  27. package/contracts/codex-app-server-0.153.4/types/TokenUsageBreakdown.ts +5 -0
  28. package/contracts/codex-app-server-0.153.4/types/TurnInterruptParams.ts +5 -0
  29. package/contracts/openhands-agent-server-v1.44.1.json +65 -0
  30. package/contracts/openhands-agent-server-v1.46.0.json +46 -0
  31. package/dist/contracts/codex-app-server-0.153.4/types/AskForApproval.d.ts +9 -0
  32. package/dist/contracts/codex-app-server-0.153.4/types/AskForApproval.js +2 -0
  33. package/dist/contracts/codex-app-server-0.153.4/types/ClientNotification.d.ts +3 -0
  34. package/dist/contracts/codex-app-server-0.153.4/types/ClientNotification.js +2 -0
  35. package/dist/contracts/codex-app-server-0.153.4/types/SandboxMode.d.ts +1 -0
  36. package/dist/contracts/codex-app-server-0.153.4/types/SandboxMode.js +2 -0
  37. package/dist/contracts/codex-app-server-0.153.4/types/ThreadReadParams.d.ts +10 -0
  38. package/dist/contracts/codex-app-server-0.153.4/types/ThreadReadParams.js +2 -0
  39. package/dist/contracts/codex-app-server-0.153.4/types/TokenUsageBreakdown.d.ts +8 -0
  40. package/dist/contracts/codex-app-server-0.153.4/types/TokenUsageBreakdown.js +2 -0
  41. package/dist/contracts/codex-app-server-0.153.4/types/TurnInterruptParams.d.ts +4 -0
  42. package/dist/contracts/codex-app-server-0.153.4/types/TurnInterruptParams.js +2 -0
  43. package/dist/src/agent-sdk-normalizer.d.ts +22 -0
  44. package/dist/src/agent-sdk-normalizer.js +995 -0
  45. package/dist/src/agent-sdk-run.d.ts +31 -0
  46. package/dist/src/agent-sdk-run.js +242 -0
  47. package/dist/src/agent-sdk-runner.d.ts +96 -0
  48. package/dist/src/agent-sdk-runner.js +407 -0
  49. package/dist/src/agent-sdk.d.ts +294 -0
  50. package/dist/src/agent-sdk.js +644 -0
  51. package/dist/src/aggregation.d.ts +183 -0
  52. package/dist/src/aggregation.js +513 -0
  53. package/dist/src/artifacts.d.ts +36 -0
  54. package/dist/src/artifacts.js +1372 -0
  55. package/dist/src/atlas-grafana.d.ts +377 -0
  56. package/dist/src/atlas-grafana.js +47 -0
  57. package/dist/src/atlas-html.d.ts +2 -0
  58. package/dist/src/atlas-html.js +34 -0
  59. package/dist/src/atlas.d.ts +130 -0
  60. package/dist/src/atlas.js +235 -0
  61. package/dist/src/behavior-assertions.d.ts +95 -0
  62. package/dist/src/behavior-assertions.js +154 -0
  63. package/dist/src/cli.d.ts +5 -0
  64. package/dist/src/cli.js +957 -0
  65. package/dist/src/codex-judge.d.ts +44 -0
  66. package/dist/src/codex-judge.js +214 -0
  67. package/dist/src/codex-run.d.ts +114 -0
  68. package/dist/src/codex-run.js +519 -0
  69. package/dist/src/codex.d.ts +187 -0
  70. package/dist/src/codex.js +1028 -0
  71. package/dist/src/contracts.d.ts +155 -0
  72. package/dist/src/contracts.js +866 -0
  73. package/dist/src/corpus.d.ts +67 -0
  74. package/dist/src/corpus.js +540 -0
  75. package/dist/src/cursor-sdk-runner.d.ts +81 -0
  76. package/dist/src/cursor-sdk-runner.js +362 -0
  77. package/dist/src/cursor-sdk.d.ts +81 -0
  78. package/dist/src/cursor-sdk.js +1078 -0
  79. package/dist/src/deepseek-adapter.d.ts +152 -0
  80. package/dist/src/deepseek-adapter.js +777 -0
  81. package/dist/src/exports.d.ts +69 -0
  82. package/dist/src/exports.js +966 -0
  83. package/dist/src/human-calibration.d.ts +167 -0
  84. package/dist/src/human-calibration.js +618 -0
  85. package/dist/src/index.d.ts +71 -0
  86. package/dist/src/index.js +38 -0
  87. package/dist/src/lifecycle.d.ts +196 -0
  88. package/dist/src/lifecycle.js +1889 -0
  89. package/dist/src/normalization-integrity.d.ts +128 -0
  90. package/dist/src/normalization-integrity.js +429 -0
  91. package/dist/src/openhands-run.d.ts +36 -0
  92. package/dist/src/openhands-run.js +415 -0
  93. package/dist/src/openhands.d.ts +130 -0
  94. package/dist/src/openhands.js +859 -0
  95. package/dist/src/pi.d.ts +203 -0
  96. package/dist/src/pi.js +1345 -0
  97. package/dist/src/process-protocol.d.ts +245 -0
  98. package/dist/src/process-protocol.js +1261 -0
  99. package/dist/src/retained-evidence.d.ts +12 -0
  100. package/dist/src/retained-evidence.js +201 -0
  101. package/dist/src/run-bundles.d.ts +193 -0
  102. package/dist/src/run-bundles.js +993 -0
  103. package/dist/src/scheduler.d.ts +109 -0
  104. package/dist/src/scheduler.js +1080 -0
  105. package/dist/src/semantic-judge.d.ts +194 -0
  106. package/dist/src/semantic-judge.js +875 -0
  107. package/dist/src/structural-observations.d.ts +124 -0
  108. package/dist/src/structural-observations.js +671 -0
  109. package/dist/src/task-packets.d.ts +115 -0
  110. package/dist/src/task-packets.js +683 -0
  111. package/dist/src/uniform-events.d.ts +120 -0
  112. package/dist/src/uniform-events.js +158 -0
  113. package/dist/src/verifiers.d.ts +116 -0
  114. package/dist/src/verifiers.js +819 -0
  115. package/dist/src/workspaces.d.ts +56 -0
  116. package/dist/src/workspaces.js +1314 -0
  117. package/docs/README.md +60 -0
  118. package/docs/development/README.md +71 -0
  119. package/docs/development/documentation-sync.md +58 -0
  120. package/docs/development/extension-contracts.md +307 -0
  121. package/docs/evaluation/README.md +48 -0
  122. package/docs/evaluation/aggregation.md +112 -0
  123. package/docs/evaluation/behavior-assertions.md +45 -0
  124. package/docs/evaluation/human-calibration.md +170 -0
  125. package/docs/evaluation/normalization-integrity.md +80 -0
  126. package/docs/evaluation/semantic-judge.md +188 -0
  127. package/docs/evaluation/structural-observations.md +84 -0
  128. package/docs/evaluation/uniform-events.md +131 -0
  129. package/docs/guides/atlas.md +144 -0
  130. package/docs/guides/evidence-and-sharing.md +100 -0
  131. package/docs/guides/operator-guide.md +424 -0
  132. package/docs/guides/quickstart.md +152 -0
  133. package/docs/guides/telemetry.md +66 -0
  134. package/docs/harnesses/README.md +31 -0
  135. package/docs/harnesses/claude-agent-sdk.md +61 -0
  136. package/docs/harnesses/codex-harness.md +174 -0
  137. package/docs/harnesses/cursor-sdk.md +149 -0
  138. package/docs/harnesses/deepseek-harness.md +134 -0
  139. package/docs/harnesses/openhands-agent-server.md +153 -0
  140. package/docs/harnesses/pi-sdk.md +66 -0
  141. package/docs/reference/README.md +19 -0
  142. package/docs/reference/agent-sdk-operational-runner.md +118 -0
  143. package/docs/reference/cli.md +114 -0
  144. package/docs/reference/contracts.md +222 -0
  145. package/docs/reference/run-bundle-contract.md +354 -0
  146. package/docs/reference/run-lifecycle.md +54 -0
  147. package/examples/cursor-sdk/README.md +11 -0
  148. package/examples/cursor-sdk/capture-profile.json +1 -0
  149. package/examples/cursor-sdk/harness.json +1 -0
  150. package/examples/cursor-sdk/model.json +1 -0
  151. package/examples/cursor-sdk/native-limits.json +1 -0
  152. package/examples/cursor-sdk/native-tool-policy.json +1 -0
  153. package/ontology/behavior-categories.v1.json +46 -0
  154. package/package.json +66 -0
  155. package/release/0.1.0/KNOWN_LIMITATIONS.md +35 -0
  156. package/release/0.1.0/README.md +46 -0
  157. package/release/0.1.0/reproducibility.json +61 -0
  158. package/release/0.2.0/KNOWN_LIMITATIONS.md +40 -0
  159. package/release/0.2.0/README.md +42 -0
  160. package/release/0.2.0/reproducibility.json +73 -0
  161. package/release/0.2.1/KNOWN_LIMITATIONS.md +18 -0
  162. package/release/0.2.1/README.md +63 -0
  163. package/release/0.2.1/reproducibility.json +73 -0
  164. package/release/README.md +13 -0
  165. package/schemas/aggregation.v1.json +293 -0
  166. package/schemas/behavior-assertions.v1.json +220 -0
  167. package/schemas/deepseek-runtime-composition.v1.schema.json +122 -0
  168. package/schemas/experiment.v1.schema.json +188 -0
  169. package/schemas/human-calibration.v1.json +336 -0
  170. package/schemas/normalization-integrity.v1.json +302 -0
  171. package/schemas/run-bundles/v1.json +1040 -0
  172. package/schemas/run-queue.v1.schema.json +206 -0
  173. package/schemas/semantic-judge.v1.json +380 -0
  174. package/schemas/structural-observations.v1.json +131 -0
  175. package/schemas/task-packet-freeze.v1.schema.json +106 -0
  176. package/schemas/task-packet.v1.schema.json +234 -0
  177. package/schemas/uniform-events/v1.json +289 -0
  178. package/scripts/atlas-grafana.sh +18 -0
@@ -0,0 +1,294 @@
1
+ import { type HookEvent, type HookInput, type Options, type PermissionMode, type SDKMessage, type SDKResultMessage } from "@anthropic-ai/claude-agent-sdk";
2
+ import type { AttemptIdentity, EvidenceSink, HarnessExecutionContext, HarnessExecutionResult, RunIdentity } from "./lifecycle.js";
3
+ export type ClaudeAgentSdkConfiguration = {
4
+ prompt: string;
5
+ model: string;
6
+ tools: string[];
7
+ allowedTools?: string[];
8
+ disallowedTools?: string[];
9
+ permissionMode: PermissionMode;
10
+ allowDangerouslySkipPermissions?: boolean;
11
+ maxTurns?: number;
12
+ maxBudgetUsd?: number;
13
+ env?: NodeJS.ProcessEnv;
14
+ telemetry?: ClaudeAgentSdkTelemetryConfiguration;
15
+ };
16
+ export type ClaudeAgentSdkTelemetrySignal = "traces" | "metrics" | "logs";
17
+ export type ClaudeAgentSdkTelemetryCorrelation = {
18
+ type: "resource-attributes";
19
+ attributes: {
20
+ "ebo.run.id": string;
21
+ "ebo.attempt.id": string;
22
+ "ebo.attempt.number": string;
23
+ };
24
+ };
25
+ export type ClaudeAgentSdkTelemetryReceipt = {
26
+ status: "received";
27
+ signals: ClaudeAgentSdkTelemetrySignal[];
28
+ } | {
29
+ status: "missing";
30
+ signals: ClaudeAgentSdkTelemetrySignal[];
31
+ reason: "not-received" | "collector-unreachable" | "process-interrupted" | "partial-receipt" | "receipt-check-failed";
32
+ } | {
33
+ status: "not-checked";
34
+ signals: [];
35
+ reason: "no-receipt-check";
36
+ };
37
+ export type ClaudeAgentSdkTelemetryConfiguration = {
38
+ endpoint: string;
39
+ protocol?: "http/protobuf" | "http/json" | "grpc";
40
+ exportIntervalMs?: number;
41
+ logUserPrompts?: boolean;
42
+ logToolDetails?: boolean;
43
+ logToolContent?: boolean;
44
+ logRawApiBodies?: boolean;
45
+ detailedHookSpans?: {
46
+ endpoint: string;
47
+ };
48
+ checkReceipt?: (correlation: ClaudeAgentSdkTelemetryCorrelation) => ClaudeAgentSdkTelemetryReceipt | Promise<ClaudeAgentSdkTelemetryReceipt>;
49
+ };
50
+ export type ClaudeAgentSdkTelemetryEvidence = {
51
+ effectiveSettings: {
52
+ endpoint: string;
53
+ protocol: "http/protobuf" | "http/json" | "grpc";
54
+ exporters: Record<ClaudeAgentSdkTelemetrySignal, "otlp">;
55
+ exportIntervalMs: Record<ClaudeAgentSdkTelemetrySignal, number>;
56
+ diagnostics: "stderr";
57
+ content: {
58
+ userPrompts: boolean;
59
+ toolDetails: boolean;
60
+ toolContent: boolean;
61
+ rawApiBodies: boolean;
62
+ };
63
+ secretHeadersConfigured: boolean;
64
+ };
65
+ correlation: ClaudeAgentSdkTelemetryCorrelation & {
66
+ inheritedResourceAttributesPreserved: boolean;
67
+ };
68
+ traces: {
69
+ status: "beta";
70
+ };
71
+ hookSpans: {
72
+ status: "disabled";
73
+ } | {
74
+ status: "enabled";
75
+ stability: "detailed-beta";
76
+ endpoint: string;
77
+ };
78
+ receipt: ClaudeAgentSdkTelemetryReceipt;
79
+ };
80
+ export type ClaudeAgentSdkEnvironment = {
81
+ env: NodeJS.ProcessEnv;
82
+ telemetry?: ClaudeAgentSdkTelemetryEvidence;
83
+ };
84
+ export type ClaudeAgentSdkLifecycleEvent = {
85
+ type: "started";
86
+ at: string;
87
+ } | {
88
+ type: "abort-requested";
89
+ at: string;
90
+ reason?: string;
91
+ } | {
92
+ type: "completed";
93
+ at: string;
94
+ messageCount: number;
95
+ } | {
96
+ type: "failed";
97
+ at: string;
98
+ messageCount: number;
99
+ error: string;
100
+ };
101
+ export type ClaudeAgentSdkMessageDiagnostic = {
102
+ sequence: number;
103
+ code: "unknown-native-type";
104
+ nativeType: string;
105
+ };
106
+ export type ClaudeAgentSdkMessageRecord = {
107
+ schemaVersion: "ebo.agent-sdk-message/v1";
108
+ sequence: number;
109
+ capturedAt: string;
110
+ nativeType: string;
111
+ nativeSubtype?: string;
112
+ sessionId?: string;
113
+ message: SDKMessage;
114
+ diagnostics?: ClaudeAgentSdkMessageDiagnostic[];
115
+ };
116
+ export type ClaudeAgentSdkCaptureReport = {
117
+ schemaVersion: "ebo.agent-sdk-capture/v1";
118
+ status: "complete" | "partial";
119
+ artifact: {
120
+ path: string;
121
+ kind: "session";
122
+ authority: "semantic";
123
+ mediaType: "application/x-ndjson";
124
+ sharingClass: "restricted";
125
+ };
126
+ capabilities: {
127
+ typedMessageStream: {
128
+ status: "available";
129
+ messageCount: number;
130
+ };
131
+ sessionIdentity: {
132
+ status: "available";
133
+ sessionId: string;
134
+ source: "sdk-result.session_id" | "sdk-message.session_id";
135
+ } | {
136
+ status: "missing";
137
+ reason: "not-emitted";
138
+ };
139
+ };
140
+ result?: {
141
+ uuid: string;
142
+ subtype: string;
143
+ sessionId: string;
144
+ };
145
+ diagnostics: {
146
+ count: number;
147
+ retained: ClaudeAgentSdkMessageDiagnostic[];
148
+ truncated: boolean;
149
+ };
150
+ };
151
+ export type ClaudeAgentSdkEvidenceSink = EvidenceSink & {
152
+ message: (message: SDKMessage) => void | Promise<void>;
153
+ stderr: (data: string) => void;
154
+ hook: (event: HookEvent, input: HookInput, toolUseId: string | undefined, signal: AbortSignal) => void | Promise<void>;
155
+ lifecycle: (event: ClaudeAgentSdkLifecycleEvent) => void | Promise<void>;
156
+ };
157
+ export type ClaudeAgentSdkHookRecord = {
158
+ schemaVersion: "ebo.claude-agent-hook/v1";
159
+ sequence: number;
160
+ callbackAt: string;
161
+ hook: HookEvent;
162
+ sessionId: string;
163
+ transcriptPath: string;
164
+ cwd: string;
165
+ promptId?: string;
166
+ toolUseId?: string;
167
+ agentId?: string;
168
+ agentType?: string;
169
+ signalAborted: boolean;
170
+ callbackOutput: Record<string, never>;
171
+ nativePayload: HookInput;
172
+ };
173
+ export type ClaudeAgentSdkHookCapture = {
174
+ path: string;
175
+ hook: ClaudeAgentSdkEvidenceSink["hook"];
176
+ flush: () => Promise<void>;
177
+ close: () => Promise<void>;
178
+ };
179
+ export type ClaudeAgentSdkCaptureWarnings = {
180
+ count: number;
181
+ diagnostic: string;
182
+ sizeBytes: number;
183
+ truncated: boolean;
184
+ };
185
+ export type ClaudeAgentSdkStreamCapture = ClaudeAgentSdkEvidenceSink & {
186
+ path: string;
187
+ flush: () => Promise<void>;
188
+ report: () => ClaudeAgentSdkCaptureReport;
189
+ close: () => Promise<void>;
190
+ };
191
+ export type ClaudeAgentSdkCapabilities = {
192
+ sdkVersion: string;
193
+ claudeCodeVersion: string;
194
+ query: {
195
+ status: "available";
196
+ };
197
+ typedMessageStream: {
198
+ status: "available";
199
+ };
200
+ stderr: {
201
+ status: "available";
202
+ };
203
+ abort: {
204
+ status: "available";
205
+ };
206
+ hookLifecycleMessages: {
207
+ status: "available";
208
+ };
209
+ hooks: Record<HookEvent, {
210
+ status: "available";
211
+ evidence: "callback";
212
+ }>;
213
+ unsupportedHooks: string[];
214
+ hookOccurrenceAuthority: "hooks.jsonl";
215
+ telemetry: {
216
+ metrics: {
217
+ status: "available";
218
+ stability: "stable";
219
+ };
220
+ logs: {
221
+ status: "available";
222
+ stability: "stable";
223
+ };
224
+ traces: {
225
+ status: "available";
226
+ stability: "beta";
227
+ };
228
+ hookSpans: {
229
+ status: "optional";
230
+ stability: "detailed-beta";
231
+ };
232
+ };
233
+ detailedBetaHookSpanTiming: {
234
+ status: "unsupported";
235
+ optional: true;
236
+ reason: string;
237
+ };
238
+ missingEvidence: Array<{
239
+ capability: "detailed-beta-hook-span-timing";
240
+ reason: string;
241
+ }>;
242
+ };
243
+ export type ClaudeAgentSdkQueryHandle = AsyncIterable<SDKMessage> & {
244
+ close: () => void;
245
+ };
246
+ export type ClaudeAgentSdkQuery = (input: {
247
+ prompt: string;
248
+ options?: Options;
249
+ }) => ClaudeAgentSdkQueryHandle;
250
+ export type ClaudeAgentSdkAttemptEvidence = {
251
+ runtime: Array<{
252
+ source: "anthropic";
253
+ name: "agent-sdk" | "agent-cli";
254
+ version: string;
255
+ }>;
256
+ effectiveConfiguration: {
257
+ model: string;
258
+ tools: string[];
259
+ allowedTools: string[];
260
+ disallowedTools: string[];
261
+ permissionMode: PermissionMode;
262
+ environment: {
263
+ parentPreserved: true;
264
+ overridesApplied: boolean;
265
+ };
266
+ budget: {
267
+ wallClockMs?: number;
268
+ maxTurns?: number;
269
+ maxBudgetUsd?: number;
270
+ };
271
+ workingDirectory: "attempt-workspace";
272
+ };
273
+ capabilities: ClaudeAgentSdkCapabilities;
274
+ telemetry?: ClaudeAgentSdkTelemetryEvidence;
275
+ usage?: {
276
+ source: "sdk-result";
277
+ totalCostUsd: number;
278
+ numTurns: number;
279
+ mainLoop: SDKResultMessage["usage"];
280
+ byModel: SDKResultMessage["modelUsage"];
281
+ };
282
+ captureWarnings: ClaudeAgentSdkCaptureWarnings;
283
+ };
284
+ export declare function openClaudeAgentSdkHookCapture(path: string, now?: () => string): Promise<ClaudeAgentSdkHookCapture>;
285
+ export declare function openClaudeAgentSdkStreamCapture(path: string, sink: ClaudeAgentSdkEvidenceSink, now?: () => string): Promise<ClaudeAgentSdkStreamCapture>;
286
+ export declare function probeClaudeAgentSdkCapabilities(): ClaudeAgentSdkCapabilities;
287
+ export declare function buildClaudeAgentSdkEnvironment(input: {
288
+ inherited: NodeJS.ProcessEnv;
289
+ overrides?: NodeJS.ProcessEnv;
290
+ telemetry?: ClaudeAgentSdkTelemetryConfiguration;
291
+ run: RunIdentity;
292
+ attempt: AttemptIdentity;
293
+ }): ClaudeAgentSdkEnvironment;
294
+ export declare function executeClaudeAgentSdk(context: HarnessExecutionContext, configuration: ClaudeAgentSdkConfiguration, sink: ClaudeAgentSdkEvidenceSink, queryFunction?: ClaudeAgentSdkQuery): Promise<HarnessExecutionResult>;