cawdex 1.35.75 → 1.35.76
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.
- package/README.md +5 -5
- package/bin/anycode.js +2 -2
- package/bin/cawdex.js +408 -408
- package/bin/ecc-hooks.cjs +11 -11
- package/dist/agents-md.d.ts +31 -0
- package/dist/agents-md.js +340 -0
- package/dist/agents-md.js.map +1 -0
- package/dist/agents.js +1424 -1424
- package/dist/api.d.ts +1 -0
- package/dist/api.js +19 -14
- package/dist/api.js.map +1 -1
- package/dist/autonomous-loops.js +287 -287
- package/dist/benchmark-repos.d.ts +31 -0
- package/dist/benchmark-repos.js +234 -8
- package/dist/benchmark-repos.js.map +1 -1
- package/dist/command-palette.js +4 -2
- package/dist/command-palette.js.map +1 -1
- package/dist/compaction.js +8 -8
- package/dist/config.js +51 -36
- package/dist/config.js.map +1 -1
- package/dist/content-engine.js +543 -543
- package/dist/context-brief.d.ts +4 -0
- package/dist/context-brief.js +230 -0
- package/dist/context-brief.js.map +1 -0
- package/dist/cost-tracker.d.ts +33 -14
- package/dist/cost-tracker.js +81 -19
- package/dist/cost-tracker.js.map +1 -1
- package/dist/coverage.js +39 -39
- package/dist/docs-sync.js +98 -98
- package/dist/evaluation.js +452 -452
- package/dist/fixed-footer.d.ts +7 -1
- package/dist/fixed-footer.js +92 -18
- package/dist/fixed-footer.js.map +1 -1
- package/dist/git-workflow.js +49 -49
- package/dist/index.d.ts +2 -0
- package/dist/index.js +161 -63
- package/dist/index.js.map +1 -1
- package/dist/live-queue.js +1 -1
- package/dist/live-queue.js.map +1 -1
- package/dist/model-aliases.d.ts +37 -0
- package/dist/model-aliases.js +203 -0
- package/dist/model-aliases.js.map +1 -0
- package/dist/orchestration.js +15 -15
- package/dist/permissions.d.ts +6 -0
- package/dist/permissions.js +53 -0
- package/dist/permissions.js.map +1 -1
- package/dist/pm2-manager.js +26 -26
- package/dist/query.d.ts +0 -1
- package/dist/query.js +74 -39
- package/dist/query.js.map +1 -1
- package/dist/refactor.js +87 -87
- package/dist/repo-command.js +7 -1
- package/dist/repo-command.js.map +1 -1
- package/dist/search-first.js +92 -92
- package/dist/skill-create.js +100 -100
- package/dist/stitch.js +1 -1
- package/dist/system-prompt.d.ts +2 -1
- package/dist/system-prompt.js +10 -5
- package/dist/system-prompt.js.map +1 -1
- package/dist/tools/github-repo-digest.d.ts +1 -1
- package/dist/tools/github-repo-digest.js +38 -6
- package/dist/tools/github-repo-digest.js.map +1 -1
- package/dist/types.d.ts +3 -0
- package/dist/types.js.map +1 -1
- package/dist/verification.js +55 -55
- package/package.json +1 -1
- package/resources/__init__.py +1 -1
- package/resources/exgentic/cawdex_agent/README.md +114 -114
- package/resources/exgentic/cawdex_agent/__init__.py +5 -5
- package/resources/exgentic/cawdex_agent/agent.py +605 -605
- package/resources/exgentic/cawdex_agent/requirements.txt +2 -2
- package/resources/exgentic/cawdex_agent/setup.sh +21 -21
- package/resources/exgentic/cawdex_agent/utils.py +1061 -1061
- package/resources/hal/cawdex_agent/README.md +24 -24
- package/resources/hal/cawdex_agent/__init__.py +1 -1
- package/resources/hal/cawdex_agent/main.py +550 -550
- package/resources/hal/cawdex_agent/requirements.txt +2 -2
- package/resources/kbench/cawdex_agent/README.md +107 -107
- package/resources/kbench/cawdex_agent/adapter.manifest.json +19 -19
- package/resources/kbench/cawdex_agent/runner.mjs +753 -753
- package/resources/open_agent_leaderboard/cawdex-agent-card.md +119 -119
- package/resources/terminal_bench/__init__.py +1 -1
- package/resources/terminal_bench/cawdex_agent.py +174 -174
- package/resources/terminal_bench/setup.sh +121 -121
|
@@ -1,753 +1,753 @@
|
|
|
1
|
-
#!/usr/bin/env node
|
|
2
|
-
import { existsSync, mkdirSync, readFileSync, readdirSync, statSync, writeFileSync } from 'node:fs';
|
|
3
|
-
import { tmpdir } from 'node:os';
|
|
4
|
-
import { dirname, join } from 'node:path';
|
|
5
|
-
import { spawnSync } from 'node:child_process';
|
|
6
|
-
|
|
7
|
-
const startedAt = Date.now();
|
|
8
|
-
|
|
9
|
-
function emit(output) {
|
|
10
|
-
process.stdout.write(JSON.stringify(output));
|
|
11
|
-
}
|
|
12
|
-
|
|
13
|
-
function fail(status, message, extra = {}) {
|
|
14
|
-
emit({
|
|
15
|
-
ok: false,
|
|
16
|
-
status,
|
|
17
|
-
failureKind: status,
|
|
18
|
-
finalText: message,
|
|
19
|
-
elapsedMs: Date.now() - startedAt,
|
|
20
|
-
artifacts: [],
|
|
21
|
-
error: { message },
|
|
22
|
-
...extra,
|
|
23
|
-
});
|
|
24
|
-
}
|
|
25
|
-
|
|
26
|
-
function readInput() {
|
|
27
|
-
const inputPath = process.env.KBENCH_ADAPTER_INPUT;
|
|
28
|
-
const raw = inputPath ? readFileSync(inputPath, 'utf8') : readFileSync(0, 'utf8');
|
|
29
|
-
return JSON.parse(raw);
|
|
30
|
-
}
|
|
31
|
-
|
|
32
|
-
function redact(text) {
|
|
33
|
-
return String(text || '')
|
|
34
|
-
.replace(/sk-or-v1-[A-Za-z0-9_-]+/g, 'sk-or-v1-[REDACTED]')
|
|
35
|
-
.replace(/sk-[A-Za-z0-9_-]{16,}/g, 'sk-[REDACTED]')
|
|
36
|
-
.replace(/hf_[A-Za-z0-9]{16,}/g, 'hf_[REDACTED]')
|
|
37
|
-
.replace(/KGAT_[A-Za-z0-9]{16,}/g, 'KGAT_[REDACTED]')
|
|
38
|
-
.replace(/npm_[A-Za-z0-9]{16,}/g, 'npm_[REDACTED]');
|
|
39
|
-
}
|
|
40
|
-
|
|
41
|
-
function truncate(text, max) {
|
|
42
|
-
const safe = redact(text);
|
|
43
|
-
if (safe.length <= max) return safe;
|
|
44
|
-
return `${safe.slice(0, max - 80)}\n...[truncated ${safe.length - (max - 80)} chars]`;
|
|
45
|
-
}
|
|
46
|
-
|
|
47
|
-
function splitCommand(command) {
|
|
48
|
-
const parts = [];
|
|
49
|
-
let cur = '';
|
|
50
|
-
let quote = null;
|
|
51
|
-
let escaped = false;
|
|
52
|
-
for (const ch of command.trim()) {
|
|
53
|
-
if (escaped) {
|
|
54
|
-
cur += ch;
|
|
55
|
-
escaped = false;
|
|
56
|
-
continue;
|
|
57
|
-
}
|
|
58
|
-
if (ch === '\\' && quote) {
|
|
59
|
-
escaped = true;
|
|
60
|
-
continue;
|
|
61
|
-
}
|
|
62
|
-
if (quote) {
|
|
63
|
-
if (ch === quote) quote = null;
|
|
64
|
-
else cur += ch;
|
|
65
|
-
continue;
|
|
66
|
-
}
|
|
67
|
-
if (ch === '"' || ch === "'") {
|
|
68
|
-
quote = ch;
|
|
69
|
-
continue;
|
|
70
|
-
}
|
|
71
|
-
if (/\s/.test(ch)) {
|
|
72
|
-
if (cur) {
|
|
73
|
-
parts.push(cur);
|
|
74
|
-
cur = '';
|
|
75
|
-
}
|
|
76
|
-
continue;
|
|
77
|
-
}
|
|
78
|
-
cur += ch;
|
|
79
|
-
}
|
|
80
|
-
if (cur) parts.push(cur);
|
|
81
|
-
return parts;
|
|
82
|
-
}
|
|
83
|
-
|
|
84
|
-
function profileForBenchmark(benchmark) {
|
|
85
|
-
const slug = String(benchmark || '').toLowerCase().replace(/[^a-z0-9]+/g, '');
|
|
86
|
-
if (slug === 'swe' || slug === 'swebench') return 'swe-bench';
|
|
87
|
-
if (slug === 'tb2' || slug === 'terminalbench') return 'terminal-bench';
|
|
88
|
-
if (slug === 'terminalworld' || slug === 'terminalworldbench' || slug === 'tw' || slug === 'tworld') return 'terminalworld';
|
|
89
|
-
if (slug === 'swechain' || slug === 'chain' || slug === 'upgrade') return 'swe-chain';
|
|
90
|
-
if (slug === 'swecycle' || slug === 'swecyclebench' || slug === 'fullcycle' || slug === 'swejudge') return 'swe-cycle';
|
|
91
|
-
if (slug === 'sweci' || slug === 'swecibench') return 'swe-ci';
|
|
92
|
-
if (slug === 'swepr' || slug === 'sweprbench' || slug === 'prbench' || slug === 'prreview' || slug === 'pullrequestreview' || slug === 'codereviewbench') return 'swe-prbench';
|
|
93
|
-
if (slug === 'tml' || slug === 'tmlbench' || slug === 'tabularml' || slug === 'kaggleml' || slug === 'kagglebench' || slug === 'datascience') return 'tml-bench';
|
|
94
|
-
if (slug === 'pi' || slug === 'pibench' || slug === 'proactive' || slug === 'proactiveassistant' || slug === 'personalassistant' || slug === 'hiddenintent') return 'pi-bench';
|
|
95
|
-
if (slug === 'cirepair' || slug === 'cirepairbench' || slug === 'ci') return 'ci-repair';
|
|
96
|
-
if (slug === 'wildclaw' || slug === 'wildclawbench' || slug === 'wcbench') return 'wildclaw';
|
|
97
|
-
if (slug === 'arc' || slug === 'arcagi' || slug === 'arcagi3' || slug === 'arcprize') return 'arc-agi';
|
|
98
|
-
if (slug === 'spec' || slug === 'specbench' || slug === 'speccompliance') return 'specbench';
|
|
99
|
-
if (slug === 'rhb' || slug === 'rewardhack' || slug === 'rewardhacking' || slug === 'rewardhackingagents') return 'reward-hacking';
|
|
100
|
-
if (slug === 'roadmap' || slug === 'roadmapbench' || slug === 'longhorizon' || slug === 'versionupgrade') return 'roadmapbench';
|
|
101
|
-
if (slug === 'saas' || slug === 'saasbench' || slug === 'enterprise') return 'saasbench';
|
|
102
|
-
if (slug === 'mobile' || slug === 'swebenchmobile' || slug === 'swemobile' || slug === 'ios') return 'swe-bench-mobile';
|
|
103
|
-
if (slug === 'webdev' || slug === 'webdevbench' || slug === 'swewebdev' || slug === 'swewebdevbench' || slug === 'vibecoding') return 'webdevbench';
|
|
104
|
-
if (slug === 'app' || slug === 'appworld' || slug === 'appworldbench') return 'appworld';
|
|
105
|
-
if (slug === 'browsecomp' || slug === 'browsecompplus' || slug === 'deepresearch' || slug === 'webresearch') return 'browsecomp';
|
|
106
|
-
if (slug === 'tau' || slug === 'tau2' || slug === 'taubench' || slug === 'taubench2' || slug.startsWith('tau2') || slug.startsWith('taubench')) return 'tau2';
|
|
107
|
-
return 'generic';
|
|
108
|
-
}
|
|
109
|
-
|
|
110
|
-
function collectTraceRefs(traceDir) {
|
|
111
|
-
const refs = [];
|
|
112
|
-
if (!traceDir || !existsSync(traceDir)) return refs;
|
|
113
|
-
const stack = [traceDir];
|
|
114
|
-
while (stack.length) {
|
|
115
|
-
const dir = stack.pop();
|
|
116
|
-
for (const entry of readdirSync(dir, { withFileTypes: true })) {
|
|
117
|
-
const full = join(dir, entry.name);
|
|
118
|
-
if (entry.isDirectory()) {
|
|
119
|
-
stack.push(full);
|
|
120
|
-
} else if (entry.name === 'summary.json' || entry.name === 'trace.jsonl') {
|
|
121
|
-
refs.push({
|
|
122
|
-
kind: entry.name === 'trace.jsonl' ? 'cawdex-tool-trace' : 'cawdex-summary',
|
|
123
|
-
path: full,
|
|
124
|
-
contentType: entry.name.endsWith('.jsonl') ? 'application/jsonl' : 'application/json',
|
|
125
|
-
description: `Cawdex ${entry.name}`,
|
|
126
|
-
});
|
|
127
|
-
}
|
|
128
|
-
}
|
|
129
|
-
}
|
|
130
|
-
return refs;
|
|
131
|
-
}
|
|
132
|
-
|
|
133
|
-
function readLatestTraceSummary(traceDir) {
|
|
134
|
-
if (!traceDir || !existsSync(traceDir)) return null;
|
|
135
|
-
const summaries = [];
|
|
136
|
-
const stack = [traceDir];
|
|
137
|
-
while (stack.length) {
|
|
138
|
-
const dir = stack.pop();
|
|
139
|
-
for (const entry of readdirSync(dir, { withFileTypes: true })) {
|
|
140
|
-
const full = join(dir, entry.name);
|
|
141
|
-
if (entry.isDirectory()) {
|
|
142
|
-
stack.push(full);
|
|
143
|
-
} else if (entry.name === 'summary.json') {
|
|
144
|
-
try {
|
|
145
|
-
const summary = JSON.parse(readFileSync(full, 'utf8'));
|
|
146
|
-
const endedAtMs = Number.isFinite(Date.parse(summary.endedAt)) ? Date.parse(summary.endedAt) : 0;
|
|
147
|
-
const mtimeMs = statSync(full).mtimeMs;
|
|
148
|
-
summaries.push({ path: full, summary, sortKey: endedAtMs || mtimeMs });
|
|
149
|
-
} catch {
|
|
150
|
-
// Ignore malformed or partially-written trace summaries.
|
|
151
|
-
}
|
|
152
|
-
}
|
|
153
|
-
}
|
|
154
|
-
}
|
|
155
|
-
summaries.sort((a, b) => b.sortKey - a.sortKey);
|
|
156
|
-
return summaries[0] || null;
|
|
157
|
-
}
|
|
158
|
-
|
|
159
|
-
function compactTraceSummary(traceSummary) {
|
|
160
|
-
if (!traceSummary) return undefined;
|
|
161
|
-
const summary = traceSummary.summary || {};
|
|
162
|
-
const quality = summary.trajectoryQuality || {};
|
|
163
|
-
const experienceCard = compactExperienceCard(summary.experienceCard);
|
|
164
|
-
return {
|
|
165
|
-
path: traceSummary.path,
|
|
166
|
-
verificationCount: summary.verificationCount,
|
|
167
|
-
verificationCommands: Array.isArray(summary.verificationCommands) ? summary.verificationCommands.slice(0, 20) : [],
|
|
168
|
-
verificationEvidence: summary.verificationEvidence,
|
|
169
|
-
finalAnswerEvidence: summary.finalAnswerEvidence,
|
|
170
|
-
usage: summary.usage,
|
|
171
|
-
experienceCard,
|
|
172
|
-
agentContextCompilation: compactAgentContextCompilation(summary.agentContextCompilation),
|
|
173
|
-
changeEvaluation: compactChangeEvaluation(summary.changeEvaluation),
|
|
174
|
-
submissionBundleManifest: compactSubmissionBundleManifest(summary.submissionBundleManifest),
|
|
175
|
-
changedFiles: Array.isArray(summary.changedFiles) ? summary.changedFiles.slice(0, 100) : [],
|
|
176
|
-
worktreeChangedFiles: Array.isArray(summary.worktreeChangedFiles) ? summary.worktreeChangedFiles.slice(0, 100) : [],
|
|
177
|
-
artifacts: Array.isArray(summary.artifacts) ? summary.artifacts.slice(0, 20) : [],
|
|
178
|
-
trajectoryQuality: {
|
|
179
|
-
benchmarkContextUsed: quality.benchmarkContextUsed,
|
|
180
|
-
usageCallCount: quality.usageCallCount,
|
|
181
|
-
usageTotalTokens: quality.usageTotalTokens,
|
|
182
|
-
usageEstimatedCostUsd: quality.usageEstimatedCostUsd,
|
|
183
|
-
costEfficiencyRisk: quality.costEfficiencyRisk,
|
|
184
|
-
totalToolElapsedMs: quality.totalToolElapsedMs,
|
|
185
|
-
maxToolElapsedMs: quality.maxToolElapsedMs,
|
|
186
|
-
slowToolCallCount: quality.slowToolCallCount,
|
|
187
|
-
slowToolEvents: Array.isArray(quality.slowToolEvents) ? quality.slowToolEvents.slice(0, 20) : [],
|
|
188
|
-
timeEfficiencyRisk: quality.timeEfficiencyRisk,
|
|
189
|
-
invalidToolActionCount: quality.invalidToolActionCount,
|
|
190
|
-
invalidToolActionPercent: quality.invalidToolActionPercent,
|
|
191
|
-
invalidToolActionEvents: Array.isArray(quality.invalidToolActionEvents) ? quality.invalidToolActionEvents.slice(0, 20) : [],
|
|
192
|
-
localizationBeforeFirstEdit: quality.localizationBeforeFirstEdit,
|
|
193
|
-
failingReproductionBeforeFirstEdit: quality.failingReproductionBeforeFirstEdit,
|
|
194
|
-
passingValidationAfterFirstEdit: quality.passingValidationAfterFirstEdit,
|
|
195
|
-
validationAfterLastEdit: quality.validationAfterLastEdit,
|
|
196
|
-
passingValidationAfterLastEdit: quality.passingValidationAfterLastEdit,
|
|
197
|
-
finalEditVerificationCount: quality.finalEditVerificationCount,
|
|
198
|
-
finalEditPassingVerificationCount: quality.finalEditPassingVerificationCount,
|
|
199
|
-
stableValidationAfterLastEdit: quality.stableValidationAfterLastEdit,
|
|
200
|
-
broadValidationAfterLastEdit: quality.broadValidationAfterLastEdit,
|
|
201
|
-
passingBroadValidationAfterLastEdit: quality.passingBroadValidationAfterLastEdit,
|
|
202
|
-
successfulVerificationCount: quality.successfulVerificationCount,
|
|
203
|
-
failedVerificationCount: quality.failedVerificationCount,
|
|
204
|
-
incompleteVerifierCount: quality.incompleteVerifierCount,
|
|
205
|
-
incompleteVerifierEvents: Array.isArray(quality.incompleteVerifierEvents) ? quality.incompleteVerifierEvents.slice(0, 20) : [],
|
|
206
|
-
inconclusiveVerifierEvents: Array.isArray(quality.inconclusiveVerifierEvents) ? quality.inconclusiveVerifierEvents.slice(0, 20) : [],
|
|
207
|
-
environmentSetupFailureCount: quality.environmentSetupFailureCount,
|
|
208
|
-
environmentSetupFailureEvents: Array.isArray(quality.environmentSetupFailureEvents) ? quality.environmentSetupFailureEvents.slice(0, 20) : [],
|
|
209
|
-
unresolvedEnvironmentSetupFailureCount: quality.unresolvedEnvironmentSetupFailureCount,
|
|
210
|
-
unresolvedEnvironmentSetupFailureEvents: Array.isArray(quality.unresolvedEnvironmentSetupFailureEvents) ? quality.unresolvedEnvironmentSetupFailureEvents.slice(0, 20) : [],
|
|
211
|
-
environmentSetupCount: quality.environmentSetupCount,
|
|
212
|
-
successfulEnvironmentSetupCount: quality.successfulEnvironmentSetupCount,
|
|
213
|
-
environmentSetupEvents: Array.isArray(quality.environmentSetupEvents) ? quality.environmentSetupEvents.slice(0, 20) : [],
|
|
214
|
-
dependencyManifestEditCount: quality.dependencyManifestEditCount,
|
|
215
|
-
dependencyLockfileEditCount: quality.dependencyLockfileEditCount,
|
|
216
|
-
dependencyManifestEditEvents: Array.isArray(quality.dependencyManifestEditEvents) ? quality.dependencyManifestEditEvents.slice(0, 20) : [],
|
|
217
|
-
dependencyLockfileEditEvents: Array.isArray(quality.dependencyLockfileEditEvents) ? quality.dependencyLockfileEditEvents.slice(0, 20) : [],
|
|
218
|
-
dependencySetupAfterManifestEdit: quality.dependencySetupAfterManifestEdit,
|
|
219
|
-
passingDependencySetupAfterManifestEdit: quality.passingDependencySetupAfterManifestEdit,
|
|
220
|
-
dependencyValidationAfterManifestEdit: quality.dependencyValidationAfterManifestEdit,
|
|
221
|
-
passingDependencyValidationAfterManifestEdit: quality.passingDependencyValidationAfterManifestEdit,
|
|
222
|
-
firstDependencySetupAfterManifestEditSeq: quality.firstDependencySetupAfterManifestEditSeq,
|
|
223
|
-
firstDependencyValidationAfterManifestEditSeq: quality.firstDependencyValidationAfterManifestEditSeq,
|
|
224
|
-
skillViewCount: quality.skillViewCount,
|
|
225
|
-
skillViewEvents: Array.isArray(quality.skillViewEvents) ? quality.skillViewEvents.slice(0, 20) : [],
|
|
226
|
-
skillNames: Array.isArray(quality.skillNames) ? quality.skillNames.slice(0, 20) : [],
|
|
227
|
-
skillLoadedBeforeLocalContext: quality.skillLoadedBeforeLocalContext,
|
|
228
|
-
excessiveSkillViewCount: quality.excessiveSkillViewCount,
|
|
229
|
-
ciWorkflowCommandCount: quality.ciWorkflowCommandCount,
|
|
230
|
-
ciVerifierCommands: Array.isArray(quality.ciVerifierCommands) ? quality.ciVerifierCommands.slice(0, 20) : [],
|
|
231
|
-
ciValidationAfterFirstEdit: quality.ciValidationAfterFirstEdit,
|
|
232
|
-
passingCiValidationAfterFirstEdit: quality.passingCiValidationAfterFirstEdit,
|
|
233
|
-
ciValidationAfterLastEdit: quality.ciValidationAfterLastEdit,
|
|
234
|
-
passingCiValidationAfterLastEdit: quality.passingCiValidationAfterLastEdit,
|
|
235
|
-
firstCiValidationAfterFirstEditSeq: quality.firstCiValidationAfterFirstEditSeq,
|
|
236
|
-
sourceResearchCoverage: quality.sourceResearchCoverage,
|
|
237
|
-
taskContractSignalCount: quality.taskContractSignalCount,
|
|
238
|
-
taskContractChecklistUsed: quality.taskContractChecklistUsed,
|
|
239
|
-
taskContractChecklistAfterContext: quality.taskContractChecklistAfterContext,
|
|
240
|
-
taskContractChecklistComplete: quality.taskContractChecklistComplete,
|
|
241
|
-
latestTodoSeq: quality.latestTodoSeq,
|
|
242
|
-
todoIncompleteCount: quality.todoIncompleteCount,
|
|
243
|
-
todoIncompleteItems: Array.isArray(quality.todoIncompleteItems) ? quality.todoIncompleteItems.slice(0, 20) : [],
|
|
244
|
-
taskAlignmentRisk: quality.taskAlignmentRisk,
|
|
245
|
-
taskAlignmentSignalCount: quality.taskAlignmentSignalCount,
|
|
246
|
-
taskAlignmentSignals: Array.isArray(quality.taskAlignmentSignals) ? quality.taskAlignmentSignals.slice(0, 20) : [],
|
|
247
|
-
specComplianceRisk: quality.specComplianceRisk,
|
|
248
|
-
specComplianceSignalCount: quality.specComplianceSignalCount,
|
|
249
|
-
specComplianceSignals: Array.isArray(quality.specComplianceSignals) ? quality.specComplianceSignals.slice(0, 20) : [],
|
|
250
|
-
rewardHackRisk: quality.rewardHackRisk,
|
|
251
|
-
rewardHackSignalCount: quality.rewardHackSignalCount,
|
|
252
|
-
rewardHackSignals: Array.isArray(quality.rewardHackSignals) ? quality.rewardHackSignals.slice(0, 20) : [],
|
|
253
|
-
longHorizonRisk: quality.longHorizonRisk,
|
|
254
|
-
longHorizonSignalCount: quality.longHorizonSignalCount,
|
|
255
|
-
longHorizonSignals: Array.isArray(quality.longHorizonSignals) ? quality.longHorizonSignals.slice(0, 20) : [],
|
|
256
|
-
proactivityDetected: quality.proactivityDetected,
|
|
257
|
-
proactivityRisk: quality.proactivityRisk,
|
|
258
|
-
proactivitySignalCount: quality.proactivitySignalCount,
|
|
259
|
-
proactivitySignals: Array.isArray(quality.proactivitySignals) ? quality.proactivitySignals.slice(0, 20) : [],
|
|
260
|
-
proactivityContextContract: quality.proactivityContextContract,
|
|
261
|
-
proactivityHiddenIntentEvidence: quality.proactivityHiddenIntentEvidence,
|
|
262
|
-
proactivityClarificationEvidence: quality.proactivityClarificationEvidence,
|
|
263
|
-
proactivityPrivacyEvidence: quality.proactivityPrivacyEvidence,
|
|
264
|
-
proactivityCompletionEvidence: quality.proactivityCompletionEvidence,
|
|
265
|
-
proactivityActionCount: quality.proactivityActionCount,
|
|
266
|
-
noEditContractDetected: quality.noEditContractDetected,
|
|
267
|
-
editAfterNoEditContract: quality.editAfterNoEditContract,
|
|
268
|
-
lastEditSeq: quality.lastEditSeq,
|
|
269
|
-
editTargetCount: quality.editTargetCount,
|
|
270
|
-
localizedEditTargetCount: quality.localizedEditTargetCount,
|
|
271
|
-
unlocalizedEditTargetEvents: Array.isArray(quality.unlocalizedEditTargetEvents) ? quality.unlocalizedEditTargetEvents.slice(0, 20) : [],
|
|
272
|
-
contextUtilizationInspectCount: quality.contextUtilizationInspectCount,
|
|
273
|
-
contextUtilizationHitCount: quality.contextUtilizationHitCount,
|
|
274
|
-
contextUtilizationMissCount: quality.contextUtilizationMissCount,
|
|
275
|
-
contextUtilizationPercent: quality.contextUtilizationPercent,
|
|
276
|
-
contextUtilizationRisk: quality.contextUtilizationRisk,
|
|
277
|
-
contextUtilizationMissEvents: Array.isArray(quality.contextUtilizationMissEvents) ? quality.contextUtilizationMissEvents.slice(0, 20) : [],
|
|
278
|
-
preEditContextInspectCount: quality.preEditContextInspectCount,
|
|
279
|
-
preEditContextHitCount: quality.preEditContextHitCount,
|
|
280
|
-
preEditContextMissCount: quality.preEditContextMissCount,
|
|
281
|
-
preEditContextUtilizationPercent: quality.preEditContextUtilizationPercent,
|
|
282
|
-
contextBloatRisk: quality.contextBloatRisk,
|
|
283
|
-
contextBloatEventCount: quality.contextBloatEventCount,
|
|
284
|
-
contextBloatEvents: Array.isArray(quality.contextBloatEvents) ? quality.contextBloatEvents.slice(0, 20) : [],
|
|
285
|
-
evidenceGroundingRisk: quality.evidenceGroundingRisk,
|
|
286
|
-
evidenceGroundingEventCount: quality.evidenceGroundingEventCount,
|
|
287
|
-
evidenceGroundingEvents: Array.isArray(quality.evidenceGroundingEvents) ? quality.evidenceGroundingEvents.slice(0, 20) : [],
|
|
288
|
-
broadEditContractDetected: quality.broadEditContractDetected,
|
|
289
|
-
largeEditSurfaceTargetCount: quality.largeEditSurfaceTargetCount,
|
|
290
|
-
largeEditSurfaceTargets: Array.isArray(quality.largeEditSurfaceTargets) ? quality.largeEditSurfaceTargets.slice(0, 40) : [],
|
|
291
|
-
redundantToolCallCount: quality.redundantToolCallCount,
|
|
292
|
-
redundantToolCallEvents: Array.isArray(quality.redundantToolCallEvents) ? quality.redundantToolCallEvents.slice(0, 20) : [],
|
|
293
|
-
redundantVerifierCount: quality.redundantVerifierCount,
|
|
294
|
-
redundantVerifierEvents: Array.isArray(quality.redundantVerifierEvents) ? quality.redundantVerifierEvents.slice(0, 20) : [],
|
|
295
|
-
blindRepairCount: quality.blindRepairCount,
|
|
296
|
-
blindRepairEvents: Array.isArray(quality.blindRepairEvents) ? quality.blindRepairEvents.slice(0, 20) : [],
|
|
297
|
-
failureAlignedRepairCount: quality.failureAlignedRepairCount,
|
|
298
|
-
failureUnalignedRepairCount: quality.failureUnalignedRepairCount,
|
|
299
|
-
failureUnalignedRepairEvents: Array.isArray(quality.failureUnalignedRepairEvents) ? quality.failureUnalignedRepairEvents.slice(0, 20) : [],
|
|
300
|
-
postEditRegressionCycleCount: quality.postEditRegressionCycleCount,
|
|
301
|
-
postEditRegressionCycleEvents: Array.isArray(quality.postEditRegressionCycleEvents) ? quality.postEditRegressionCycleEvents.slice(0, 20) : [],
|
|
302
|
-
postSuccessMutationCount: quality.postSuccessMutationCount,
|
|
303
|
-
postSuccessMutationEvents: Array.isArray(quality.postSuccessMutationEvents) ? quality.postSuccessMutationEvents.slice(0, 20) : [],
|
|
304
|
-
predictedEditCount: quality.predictedEditCount,
|
|
305
|
-
unpredictedEditCount: quality.unpredictedEditCount,
|
|
306
|
-
contradictedEditPredictionCount: quality.contradictedEditPredictionCount,
|
|
307
|
-
unverifiedEditPredictionCount: quality.unverifiedEditPredictionCount,
|
|
308
|
-
decisionObservabilityRisk: quality.decisionObservabilityRisk,
|
|
309
|
-
scratchArtifactPermissionDetected: quality.scratchArtifactPermissionDetected,
|
|
310
|
-
scratchArtifactEvents: Array.isArray(quality.scratchArtifactEvents) ? quality.scratchArtifactEvents.slice(0, 20) : [],
|
|
311
|
-
postEditDiffReview: quality.postEditDiffReview,
|
|
312
|
-
diffReviewAfterLastEdit: quality.diffReviewAfterLastEdit,
|
|
313
|
-
firstPostEditDiffReviewSeq: quality.firstPostEditDiffReviewSeq,
|
|
314
|
-
firstDiffReviewAfterLastEditSeq: quality.firstDiffReviewAfterLastEditSeq,
|
|
315
|
-
broadValidationAfterFirstEdit: quality.broadValidationAfterFirstEdit,
|
|
316
|
-
passingBroadValidationAfterFirstEdit: quality.passingBroadValidationAfterFirstEdit,
|
|
317
|
-
firstBroadValidationAfterFirstEditSeq: quality.firstBroadValidationAfterFirstEditSeq,
|
|
318
|
-
lastPostEditVerificationSeq: quality.lastPostEditVerificationSeq,
|
|
319
|
-
lastPostEditVerificationStatus: quality.lastPostEditVerificationStatus,
|
|
320
|
-
lastPostEditVerificationConclusiveFailure: quality.lastPostEditVerificationConclusiveFailure,
|
|
321
|
-
firstConclusiveFailedVerificationSeq: quality.firstConclusiveFailedVerificationSeq,
|
|
322
|
-
testEditPermissionDetected: quality.testEditPermissionDetected,
|
|
323
|
-
testHarnessEditEvents: Array.isArray(quality.testHarnessEditEvents) ? quality.testHarnessEditEvents.slice(0, 20) : [],
|
|
324
|
-
processScore: quality.processScore,
|
|
325
|
-
processDefects: Array.isArray(quality.processDefects) ? quality.processDefects.slice(0, 20) : [],
|
|
326
|
-
warnings: Array.isArray(quality.warnings) ? quality.warnings.slice(0, 20) : [],
|
|
327
|
-
},
|
|
328
|
-
};
|
|
329
|
-
}
|
|
330
|
-
|
|
331
|
-
function compactAgentContextCompilation(compilation) {
|
|
332
|
-
if (!compilation || typeof compilation !== 'object' || Array.isArray(compilation)) return undefined;
|
|
333
|
-
const metadata = compilation.metadata && typeof compilation.metadata === 'object' && !Array.isArray(compilation.metadata)
|
|
334
|
-
? compilation.metadata
|
|
335
|
-
: {};
|
|
336
|
-
return {
|
|
337
|
-
version: compilation.version,
|
|
338
|
-
format: compilation.format,
|
|
339
|
-
task: truncate(compilation.task || '', 2000),
|
|
340
|
-
context: truncate(compilation.context || '', 5000),
|
|
341
|
-
answer: truncate(compilation.answer || '', 2500),
|
|
342
|
-
metadata: {
|
|
343
|
-
sessionId: metadata.sessionId,
|
|
344
|
-
mode: metadata.mode,
|
|
345
|
-
provider: metadata.provider,
|
|
346
|
-
model: metadata.model,
|
|
347
|
-
eventCount: metadata.eventCount,
|
|
348
|
-
contextEventCount: metadata.contextEventCount,
|
|
349
|
-
verificationStatus: metadata.verificationStatus,
|
|
350
|
-
successfulVerificationCount: metadata.successfulVerificationCount,
|
|
351
|
-
processScore: metadata.processScore,
|
|
352
|
-
usageTotalTokens: metadata.usageTotalTokens,
|
|
353
|
-
estimatedCostUsd: metadata.estimatedCostUsd,
|
|
354
|
-
changedFiles: Array.isArray(metadata.changedFiles) ? metadata.changedFiles.slice(0, 100) : [],
|
|
355
|
-
verificationCommands: Array.isArray(metadata.verificationCommands) ? metadata.verificationCommands.slice(0, 20) : [],
|
|
356
|
-
sourceResearchCoverage: metadata.sourceResearchCoverage,
|
|
357
|
-
warnings: Array.isArray(metadata.warnings) ? metadata.warnings.slice(0, 20) : [],
|
|
358
|
-
},
|
|
359
|
-
};
|
|
360
|
-
}
|
|
361
|
-
|
|
362
|
-
function compactChangeEvaluation(changeEvaluation) {
|
|
363
|
-
if (!changeEvaluation || typeof changeEvaluation !== 'object' || Array.isArray(changeEvaluation)) return undefined;
|
|
364
|
-
return {
|
|
365
|
-
version: changeEvaluation.version,
|
|
366
|
-
format: changeEvaluation.format,
|
|
367
|
-
status: changeEvaluation.status,
|
|
368
|
-
accepted: changeEvaluation.accepted,
|
|
369
|
-
reason: truncate(changeEvaluation.reason || '', 500),
|
|
370
|
-
editCount: changeEvaluation.editCount,
|
|
371
|
-
predictedEditCount: changeEvaluation.predictedEditCount,
|
|
372
|
-
unpredictedEditCount: changeEvaluation.unpredictedEditCount,
|
|
373
|
-
confirmedPredictionCount: changeEvaluation.confirmedPredictionCount,
|
|
374
|
-
contradictedPredictionCount: changeEvaluation.contradictedPredictionCount,
|
|
375
|
-
unverifiedPredictionCount: changeEvaluation.unverifiedPredictionCount,
|
|
376
|
-
decisionCoveragePercent: changeEvaluation.decisionCoveragePercent,
|
|
377
|
-
regressionCycleCount: changeEvaluation.regressionCycleCount,
|
|
378
|
-
broadRegressionFailureCount: changeEvaluation.broadRegressionFailureCount,
|
|
379
|
-
predictions: Array.isArray(changeEvaluation.predictions) ? changeEvaluation.predictions.slice(0, 20) : [],
|
|
380
|
-
unpredictedEdits: Array.isArray(changeEvaluation.unpredictedEdits) ? changeEvaluation.unpredictedEdits.slice(0, 20) : [],
|
|
381
|
-
regressionCycles: Array.isArray(changeEvaluation.regressionCycles) ? changeEvaluation.regressionCycles.slice(0, 20) : [],
|
|
382
|
-
recommendedAction: truncate(changeEvaluation.recommendedAction || '', 500),
|
|
383
|
-
};
|
|
384
|
-
}
|
|
385
|
-
|
|
386
|
-
function compactSubmissionBundleManifest(manifest) {
|
|
387
|
-
if (!manifest || typeof manifest !== 'object' || Array.isArray(manifest)) return undefined;
|
|
388
|
-
return {
|
|
389
|
-
version: manifest.version,
|
|
390
|
-
format: manifest.format,
|
|
391
|
-
submissionReady: manifest.submissionReady,
|
|
392
|
-
reason: truncate(manifest.reason || '', 500),
|
|
393
|
-
officialResultRequired: manifest.officialResultRequired,
|
|
394
|
-
missingOfficialFields: Array.isArray(manifest.missingOfficialFields) ? manifest.missingOfficialFields.slice(0, 20) : [],
|
|
395
|
-
benchmark: manifest.benchmark,
|
|
396
|
-
benchmarkName: manifest.benchmarkName,
|
|
397
|
-
sessionId: manifest.sessionId,
|
|
398
|
-
provider: manifest.provider,
|
|
399
|
-
model: manifest.model,
|
|
400
|
-
summaryContainer: manifest.summaryContainer,
|
|
401
|
-
artifacts: Array.isArray(manifest.artifacts) ? manifest.artifacts.slice(0, 50) : [],
|
|
402
|
-
verification: manifest.verification,
|
|
403
|
-
usage: manifest.usage,
|
|
404
|
-
process: manifest.process,
|
|
405
|
-
leaderboardDraft: manifest.leaderboardDraft,
|
|
406
|
-
};
|
|
407
|
-
}
|
|
408
|
-
|
|
409
|
-
function compactExperienceCard(card) {
|
|
410
|
-
if (!card || typeof card !== 'object' || Array.isArray(card)) return undefined;
|
|
411
|
-
return {
|
|
412
|
-
version: card.version,
|
|
413
|
-
replayCheckpoints: Array.isArray(card.replayCheckpoints) ? card.replayCheckpoints.slice(0, 20) : [],
|
|
414
|
-
failureSignatures: Array.isArray(card.failureSignatures) ? card.failureSignatures.slice(0, 10) : [],
|
|
415
|
-
sourceResearchCoverage: card.sourceResearchCoverage,
|
|
416
|
-
taskContract: card.taskContract,
|
|
417
|
-
taskAlignment: compactRiskSignalBlock(card.taskAlignment),
|
|
418
|
-
specCompliance: compactRiskSignalBlock(card.specCompliance),
|
|
419
|
-
rewardHack: compactRiskSignalBlock(card.rewardHack),
|
|
420
|
-
longHorizon: compactRiskSignalBlock(card.longHorizon),
|
|
421
|
-
proactivity: compactProactivity(card.proactivity),
|
|
422
|
-
environmentReconstruction: compactEnvironmentReconstruction(card.environmentReconstruction),
|
|
423
|
-
dependencyUpgrade: compactDependencyUpgrade(card.dependencyUpgrade),
|
|
424
|
-
decisionObservability: compactDecisionObservability(card.decisionObservability),
|
|
425
|
-
validationReliability: compactValidationReliability(card.validationReliability),
|
|
426
|
-
contextUtilization: compactContextUtilization(card.contextUtilization),
|
|
427
|
-
runEfficiency: compactRunEfficiency(card.runEfficiency),
|
|
428
|
-
verificationCommands: Array.isArray(card.verificationCommands) ? card.verificationCommands.slice(0, 20) : [],
|
|
429
|
-
changedFiles: Array.isArray(card.changedFiles) ? card.changedFiles.slice(0, 100) : [],
|
|
430
|
-
warnings: Array.isArray(card.warnings) ? card.warnings.slice(0, 20) : [],
|
|
431
|
-
};
|
|
432
|
-
}
|
|
433
|
-
|
|
434
|
-
function compactRiskSignalBlock(block) {
|
|
435
|
-
if (!block || typeof block !== 'object' || Array.isArray(block)) return undefined;
|
|
436
|
-
return {
|
|
437
|
-
risk: block.risk,
|
|
438
|
-
signalCount: block.signalCount,
|
|
439
|
-
signals: Array.isArray(block.signals) ? block.signals.slice(0, 20) : [],
|
|
440
|
-
};
|
|
441
|
-
}
|
|
442
|
-
|
|
443
|
-
function compactProactivity(proactivity) {
|
|
444
|
-
if (!proactivity || typeof proactivity !== 'object' || Array.isArray(proactivity)) return undefined;
|
|
445
|
-
return {
|
|
446
|
-
detected: proactivity.detected,
|
|
447
|
-
risk: proactivity.risk,
|
|
448
|
-
signalCount: proactivity.signalCount,
|
|
449
|
-
signals: Array.isArray(proactivity.signals) ? proactivity.signals.slice(0, 20) : [],
|
|
450
|
-
contextContract: proactivity.contextContract,
|
|
451
|
-
hiddenIntentEvidence: proactivity.hiddenIntentEvidence,
|
|
452
|
-
clarificationEvidence: proactivity.clarificationEvidence,
|
|
453
|
-
privacyEvidence: proactivity.privacyEvidence,
|
|
454
|
-
completionEvidence: proactivity.completionEvidence,
|
|
455
|
-
actionCount: proactivity.actionCount,
|
|
456
|
-
};
|
|
457
|
-
}
|
|
458
|
-
|
|
459
|
-
function compactRunEfficiency(runEfficiency) {
|
|
460
|
-
if (!runEfficiency || typeof runEfficiency !== 'object' || Array.isArray(runEfficiency)) return undefined;
|
|
461
|
-
return {
|
|
462
|
-
toolCallCount: runEfficiency.toolCallCount,
|
|
463
|
-
totalToolElapsedMs: runEfficiency.totalToolElapsedMs,
|
|
464
|
-
maxToolElapsedMs: runEfficiency.maxToolElapsedMs,
|
|
465
|
-
slowToolCallCount: runEfficiency.slowToolCallCount,
|
|
466
|
-
usageCallCount: runEfficiency.usageCallCount,
|
|
467
|
-
totalTokens: runEfficiency.totalTokens,
|
|
468
|
-
estimatedCostUsd: runEfficiency.estimatedCostUsd,
|
|
469
|
-
successfulVerificationCount: runEfficiency.successfulVerificationCount,
|
|
470
|
-
processScore: runEfficiency.processScore,
|
|
471
|
-
processDefectCount: runEfficiency.processDefectCount,
|
|
472
|
-
warningCount: runEfficiency.warningCount,
|
|
473
|
-
invalidToolActionCount: runEfficiency.invalidToolActionCount,
|
|
474
|
-
invalidToolActionPercent: runEfficiency.invalidToolActionPercent,
|
|
475
|
-
costEfficiencyRisk: runEfficiency.costEfficiencyRisk,
|
|
476
|
-
timeEfficiencyRisk: runEfficiency.timeEfficiencyRisk,
|
|
477
|
-
slowToolEvents: Array.isArray(runEfficiency.slowToolEvents) ? runEfficiency.slowToolEvents.slice(0, 12) : [],
|
|
478
|
-
};
|
|
479
|
-
}
|
|
480
|
-
|
|
481
|
-
function compactContextUtilization(contextUtilization) {
|
|
482
|
-
if (!contextUtilization || typeof contextUtilization !== 'object' || Array.isArray(contextUtilization)) return undefined;
|
|
483
|
-
return {
|
|
484
|
-
inspectCount: contextUtilization.inspectCount,
|
|
485
|
-
hitCount: contextUtilization.hitCount,
|
|
486
|
-
missCount: contextUtilization.missCount,
|
|
487
|
-
utilizationPercent: contextUtilization.utilizationPercent,
|
|
488
|
-
risk: contextUtilization.risk,
|
|
489
|
-
missEvents: Array.isArray(contextUtilization.missEvents) ? contextUtilization.missEvents.slice(0, 12) : [],
|
|
490
|
-
};
|
|
491
|
-
}
|
|
492
|
-
|
|
493
|
-
function compactValidationReliability(validationReliability) {
|
|
494
|
-
if (!validationReliability || typeof validationReliability !== 'object' || Array.isArray(validationReliability)) return undefined;
|
|
495
|
-
return {
|
|
496
|
-
lastEditSeq: validationReliability.lastEditSeq,
|
|
497
|
-
finalEditVerificationCount: validationReliability.finalEditVerificationCount,
|
|
498
|
-
finalEditPassingVerificationCount: validationReliability.finalEditPassingVerificationCount,
|
|
499
|
-
stableValidationAfterLastEdit: validationReliability.stableValidationAfterLastEdit,
|
|
500
|
-
broadValidationAfterLastEdit: validationReliability.broadValidationAfterLastEdit,
|
|
501
|
-
passingBroadValidationAfterLastEdit: validationReliability.passingBroadValidationAfterLastEdit,
|
|
502
|
-
ciValidationAfterLastEdit: validationReliability.ciValidationAfterLastEdit,
|
|
503
|
-
passingCiValidationAfterLastEdit: validationReliability.passingCiValidationAfterLastEdit,
|
|
504
|
-
postEditRegressionCycleCount: validationReliability.postEditRegressionCycleCount,
|
|
505
|
-
lastPostEditVerificationSeq: validationReliability.lastPostEditVerificationSeq,
|
|
506
|
-
lastPostEditVerificationStatus: validationReliability.lastPostEditVerificationStatus,
|
|
507
|
-
finalVerifierCommands: Array.isArray(validationReliability.finalVerifierCommands) ? validationReliability.finalVerifierCommands.slice(0, 12) : [],
|
|
508
|
-
};
|
|
509
|
-
}
|
|
510
|
-
|
|
511
|
-
function compactDecisionObservability(decisionObservability) {
|
|
512
|
-
if (!decisionObservability || typeof decisionObservability !== 'object' || Array.isArray(decisionObservability)) return undefined;
|
|
513
|
-
return {
|
|
514
|
-
editCount: decisionObservability.editCount,
|
|
515
|
-
predictedEditCount: decisionObservability.predictedEditCount,
|
|
516
|
-
verifiedPredictionCount: decisionObservability.verifiedPredictionCount,
|
|
517
|
-
editPredictions: Array.isArray(decisionObservability.editPredictions) ? decisionObservability.editPredictions.slice(0, 12) : [],
|
|
518
|
-
};
|
|
519
|
-
}
|
|
520
|
-
|
|
521
|
-
function compactEnvironmentReconstruction(environmentReconstruction) {
|
|
522
|
-
if (!environmentReconstruction || typeof environmentReconstruction !== 'object' || Array.isArray(environmentReconstruction)) return undefined;
|
|
523
|
-
return {
|
|
524
|
-
setupFailureCount: environmentReconstruction.setupFailureCount,
|
|
525
|
-
unresolvedSetupFailureCount: environmentReconstruction.unresolvedSetupFailureCount,
|
|
526
|
-
setupCount: environmentReconstruction.setupCount,
|
|
527
|
-
successfulSetupCount: environmentReconstruction.successfulSetupCount,
|
|
528
|
-
setupEvents: Array.isArray(environmentReconstruction.setupEvents) ? environmentReconstruction.setupEvents.slice(0, 12) : [],
|
|
529
|
-
setupFailures: Array.isArray(environmentReconstruction.setupFailures) ? environmentReconstruction.setupFailures.slice(0, 12) : [],
|
|
530
|
-
unresolvedSetupFailures: Array.isArray(environmentReconstruction.unresolvedSetupFailures) ? environmentReconstruction.unresolvedSetupFailures.slice(0, 12) : [],
|
|
531
|
-
};
|
|
532
|
-
}
|
|
533
|
-
|
|
534
|
-
function compactDependencyUpgrade(dependencyUpgrade) {
|
|
535
|
-
if (!dependencyUpgrade || typeof dependencyUpgrade !== 'object' || Array.isArray(dependencyUpgrade)) return undefined;
|
|
536
|
-
return {
|
|
537
|
-
manifestEditCount: dependencyUpgrade.manifestEditCount,
|
|
538
|
-
lockfileEditCount: dependencyUpgrade.lockfileEditCount,
|
|
539
|
-
manifestEdits: Array.isArray(dependencyUpgrade.manifestEdits) ? dependencyUpgrade.manifestEdits.slice(0, 12) : [],
|
|
540
|
-
lockfileEdits: Array.isArray(dependencyUpgrade.lockfileEdits) ? dependencyUpgrade.lockfileEdits.slice(0, 12) : [],
|
|
541
|
-
setupAfterManifestEdit: dependencyUpgrade.setupAfterManifestEdit,
|
|
542
|
-
passingSetupAfterManifestEdit: dependencyUpgrade.passingSetupAfterManifestEdit,
|
|
543
|
-
validationAfterManifestEdit: dependencyUpgrade.validationAfterManifestEdit,
|
|
544
|
-
passingValidationAfterManifestEdit: dependencyUpgrade.passingValidationAfterManifestEdit,
|
|
545
|
-
firstSetupAfterManifestEditSeq: dependencyUpgrade.firstSetupAfterManifestEditSeq,
|
|
546
|
-
firstValidationAfterManifestEditSeq: dependencyUpgrade.firstValidationAfterManifestEditSeq,
|
|
547
|
-
};
|
|
548
|
-
}
|
|
549
|
-
|
|
550
|
-
function collectGitArtifactRefs(workdir, artifactRoot) {
|
|
551
|
-
const refs = [];
|
|
552
|
-
if (!workdir || !existsSync(workdir)) return refs;
|
|
553
|
-
|
|
554
|
-
const gitCheck = spawnSync('git', ['-C', workdir, 'rev-parse', '--is-inside-work-tree'], {
|
|
555
|
-
encoding: 'utf8',
|
|
556
|
-
timeout: 5000,
|
|
557
|
-
maxBuffer: 64 * 1024,
|
|
558
|
-
});
|
|
559
|
-
if (gitCheck.status !== 0 || !String(gitCheck.stdout || '').trim().startsWith('true')) {
|
|
560
|
-
return refs;
|
|
561
|
-
}
|
|
562
|
-
|
|
563
|
-
const diff = buildWorktreePatch(workdir);
|
|
564
|
-
if (diff.trim()) {
|
|
565
|
-
const patchPath = join(artifactRoot, 'cawdex.patch');
|
|
566
|
-
writeFileSync(patchPath, redact(diff), 'utf8');
|
|
567
|
-
refs.push({
|
|
568
|
-
kind: 'patch',
|
|
569
|
-
path: patchPath,
|
|
570
|
-
contentType: 'text/x-diff',
|
|
571
|
-
description: 'Redacted git diff after Cawdex run.',
|
|
572
|
-
});
|
|
573
|
-
}
|
|
574
|
-
|
|
575
|
-
const status = spawnSync('git', ['-C', workdir, 'status', '--short'], {
|
|
576
|
-
encoding: 'utf8',
|
|
577
|
-
timeout: 5000,
|
|
578
|
-
maxBuffer: 512 * 1024,
|
|
579
|
-
});
|
|
580
|
-
if (status.stdout && status.stdout.trim()) {
|
|
581
|
-
const statusPath = join(artifactRoot, 'git-status.txt');
|
|
582
|
-
writeFileSync(statusPath, redact(status.stdout), 'utf8');
|
|
583
|
-
refs.push({
|
|
584
|
-
kind: 'git-status',
|
|
585
|
-
path: statusPath,
|
|
586
|
-
contentType: 'text/plain',
|
|
587
|
-
description: 'Redacted git status after Cawdex run.',
|
|
588
|
-
});
|
|
589
|
-
}
|
|
590
|
-
|
|
591
|
-
return refs;
|
|
592
|
-
}
|
|
593
|
-
|
|
594
|
-
function runGit(workdir, args, options = {}) {
|
|
595
|
-
const result = spawnSync('git', ['-C', workdir, ...args], {
|
|
596
|
-
encoding: 'utf8',
|
|
597
|
-
timeout: options.timeout || 5000,
|
|
598
|
-
maxBuffer: options.maxBuffer || 1024 * 1024,
|
|
599
|
-
});
|
|
600
|
-
if (result.error) return '';
|
|
601
|
-
if (result.status !== 0 && !(options.allowDiffExit && result.status === 1)) return '';
|
|
602
|
-
return result.stdout || '';
|
|
603
|
-
}
|
|
604
|
-
|
|
605
|
-
function buildWorktreePatch(workdir) {
|
|
606
|
-
const parts = [
|
|
607
|
-
runGit(workdir, ['diff', '--binary', '--no-ext-diff'], { timeout: 10000, maxBuffer: 20 * 1024 * 1024 }),
|
|
608
|
-
runGit(workdir, ['diff', '--cached', '--binary', '--no-ext-diff'], { timeout: 10000, maxBuffer: 20 * 1024 * 1024 }),
|
|
609
|
-
...collectUntrackedDiffs(workdir),
|
|
610
|
-
].map((part) => part.trim()).filter(Boolean);
|
|
611
|
-
return parts.join('\n\n') + (parts.length ? '\n' : '');
|
|
612
|
-
}
|
|
613
|
-
|
|
614
|
-
function collectUntrackedDiffs(workdir) {
|
|
615
|
-
const raw = runGit(workdir, ['ls-files', '--others', '--exclude-standard', '-z']);
|
|
616
|
-
const files = raw.split('\0').map((file) => file.trim()).filter(Boolean).slice(0, 80);
|
|
617
|
-
return files
|
|
618
|
-
.map((file) => runGit(workdir, ['diff', '--no-index', '--binary', '--no-ext-diff', '--', '/dev/null', file], {
|
|
619
|
-
timeout: 5000,
|
|
620
|
-
maxBuffer: 5 * 1024 * 1024,
|
|
621
|
-
allowDiffExit: true,
|
|
622
|
-
}))
|
|
623
|
-
.filter((diff) => diff.trim());
|
|
624
|
-
}
|
|
625
|
-
|
|
626
|
-
let payload;
|
|
627
|
-
try {
|
|
628
|
-
payload = readInput();
|
|
629
|
-
} catch (err) {
|
|
630
|
-
fail('invalid_adapter', `Could not read adapter input: ${err?.message || err}`);
|
|
631
|
-
process.exit(0);
|
|
632
|
-
}
|
|
633
|
-
|
|
634
|
-
if (payload.mode !== 'task') {
|
|
635
|
-
fail('unsupported_capability', 'Cawdex KBench adapter currently supports task mode only.');
|
|
636
|
-
process.exit(0);
|
|
637
|
-
}
|
|
638
|
-
|
|
639
|
-
const task = payload.task || {};
|
|
640
|
-
const config = payload.config || {};
|
|
641
|
-
const env = payload.env || task.env || {};
|
|
642
|
-
const taskEnv = task.env || {};
|
|
643
|
-
const benchmark = task.benchmark || 'swe';
|
|
644
|
-
const instruction = String(task.instruction || '').trim();
|
|
645
|
-
const profile = profileForBenchmark(benchmark);
|
|
646
|
-
const prompt = `/benchmark ${profile} ${instruction}`;
|
|
647
|
-
const workdir = config.workDir || env.workdir || env.repoPath || taskEnv.workdir || taskEnv.repoPath || process.cwd();
|
|
648
|
-
const artifactRoot = config.storeDir
|
|
649
|
-
|| process.env.CAWDEX_KBENCH_ARTIFACT_DIR
|
|
650
|
-
|| (() => {
|
|
651
|
-
const dir = join(tmpdir(), `cawdex-kbench-${process.pid}-${Date.now()}`);
|
|
652
|
-
mkdirSync(dir, { recursive: true });
|
|
653
|
-
return dir;
|
|
654
|
-
})();
|
|
655
|
-
mkdirSync(artifactRoot, { recursive: true });
|
|
656
|
-
|
|
657
|
-
const stdoutPath = join(artifactRoot, 'cawdex.stdout.txt');
|
|
658
|
-
const stderrPath = join(artifactRoot, 'cawdex.stderr.txt');
|
|
659
|
-
const instructionPath = join(artifactRoot, 'instruction.txt');
|
|
660
|
-
const traceDir = join(artifactRoot, 'cawdex-trace');
|
|
661
|
-
mkdirSync(dirname(stdoutPath), { recursive: true });
|
|
662
|
-
writeFileSync(instructionPath, redact(instruction), 'utf8');
|
|
663
|
-
|
|
664
|
-
const commandParts = splitCommand(process.env.CAWDEX_KBENCH_COMMAND || process.env.CAWDEX_KBENCH_COMMAND || 'cawdex');
|
|
665
|
-
if (!commandParts.length) {
|
|
666
|
-
fail('invalid_adapter', 'CAWDEX_KBENCH_COMMAND resolved to an empty command.');
|
|
667
|
-
process.exit(0);
|
|
668
|
-
}
|
|
669
|
-
|
|
670
|
-
const [command, ...prefixArgs] = commandParts;
|
|
671
|
-
const args = [
|
|
672
|
-
...prefixArgs,
|
|
673
|
-
'--prompt', prompt,
|
|
674
|
-
'--perm', process.env.CAWDEX_KBENCH_PERMISSION || 'yolo',
|
|
675
|
-
'--output-format', 'text',
|
|
676
|
-
'--benchmark-trace-dir', traceDir,
|
|
677
|
-
];
|
|
678
|
-
if (config.modelName) args.push('--model', String(config.modelName));
|
|
679
|
-
if (config.temperature !== undefined) args.push('--temperature', String(config.temperature));
|
|
680
|
-
if (config.baseUrl) args.push('--base-url', String(config.baseUrl));
|
|
681
|
-
if (config.apiKeyEnv) args.push('--api-key-env', String(config.apiKeyEnv));
|
|
682
|
-
if (process.env.CAWDEX_KBENCH_EXTRA_ARGS) {
|
|
683
|
-
args.push(...splitCommand(process.env.CAWDEX_KBENCH_EXTRA_ARGS));
|
|
684
|
-
}
|
|
685
|
-
|
|
686
|
-
const childEnv = {
|
|
687
|
-
...process.env,
|
|
688
|
-
CAWDEX_BENCHMARK_TRACE: '1',
|
|
689
|
-
CAWDEX_BENCHMARK_TRACE_DIR: traceDir,
|
|
690
|
-
CAWDEX_BASH_TIMEOUT_MS: process.env.CAWDEX_BASH_TIMEOUT_MS || '300000',
|
|
691
|
-
};
|
|
692
|
-
for (const [key, value] of Object.entries(env.envVars || {})) {
|
|
693
|
-
if (typeof value === 'string') childEnv[key] = value;
|
|
694
|
-
}
|
|
695
|
-
for (const [key, value] of Object.entries(taskEnv.envVars || {})) {
|
|
696
|
-
if (typeof value === 'string') childEnv[key] = value;
|
|
697
|
-
}
|
|
698
|
-
|
|
699
|
-
const result = spawnSync(command, args, {
|
|
700
|
-
cwd: existsSync(workdir) ? workdir : process.cwd(),
|
|
701
|
-
env: childEnv,
|
|
702
|
-
encoding: 'utf8',
|
|
703
|
-
timeout: typeof config.timeoutMs === 'number' && config.timeoutMs > 0 ? config.timeoutMs : undefined,
|
|
704
|
-
maxBuffer: 20 * 1024 * 1024,
|
|
705
|
-
});
|
|
706
|
-
|
|
707
|
-
const stdout = result.stdout || '';
|
|
708
|
-
const stderr = result.stderr || '';
|
|
709
|
-
writeFileSync(stdoutPath, redact(stdout), 'utf8');
|
|
710
|
-
writeFileSync(stderrPath, redact(stderr), 'utf8');
|
|
711
|
-
const elapsedMs = Date.now() - startedAt;
|
|
712
|
-
const timedOut = result.error?.code === 'ETIMEDOUT' || result.signal === 'SIGTERM';
|
|
713
|
-
const exitCode = typeof result.status === 'number' ? result.status : (timedOut ? 124 : 1);
|
|
714
|
-
const ok = exitCode === 0;
|
|
715
|
-
const traceRefs = collectTraceRefs(traceDir);
|
|
716
|
-
const traceSummary = compactTraceSummary(readLatestTraceSummary(traceDir));
|
|
717
|
-
const workdirUsed = existsSync(workdir) ? workdir : process.cwd();
|
|
718
|
-
const gitRefs = collectGitArtifactRefs(workdirUsed, artifactRoot);
|
|
719
|
-
const stdoutLines = stdout.split(/\r?\n/).map((line) => line.trim()).filter(Boolean);
|
|
720
|
-
const finalText = stdoutLines.at(-1) || stdout.trim() || (ok ? 'Cawdex completed.' : 'Cawdex produced no stdout.');
|
|
721
|
-
const artifacts = [
|
|
722
|
-
{ kind: 'instruction', path: instructionPath, contentType: 'text/plain', description: 'KBench task instruction passed to Cawdex.' },
|
|
723
|
-
{ kind: 'stdout', path: stdoutPath, contentType: 'text/plain', description: 'Cawdex stdout.' },
|
|
724
|
-
{ kind: 'stderr', path: stderrPath, contentType: 'text/plain', description: 'Cawdex stderr.' },
|
|
725
|
-
...gitRefs,
|
|
726
|
-
...traceRefs,
|
|
727
|
-
];
|
|
728
|
-
|
|
729
|
-
const output = {
|
|
730
|
-
ok,
|
|
731
|
-
status: ok ? 'ok' : (timedOut ? 'timeout' : 'agent_error'),
|
|
732
|
-
failureKind: ok ? undefined : (timedOut ? 'timeout' : `exit_${exitCode}`),
|
|
733
|
-
finalText: truncate(finalText, 4000),
|
|
734
|
-
elapsedMs,
|
|
735
|
-
artifacts,
|
|
736
|
-
trace: traceRefs.length ? { native: traceRefs } : undefined,
|
|
737
|
-
benchmarkResult: {
|
|
738
|
-
mode: 'cawdex-kbench',
|
|
739
|
-
benchmark,
|
|
740
|
-
profile,
|
|
741
|
-
exitCode,
|
|
742
|
-
workdir: workdirUsed,
|
|
743
|
-
traceSummary,
|
|
744
|
-
verificationEvidence: traceSummary?.verificationEvidence,
|
|
745
|
-
experienceCard: traceSummary?.experienceCard,
|
|
746
|
-
usage: traceSummary?.usage,
|
|
747
|
-
},
|
|
748
|
-
error: ok ? undefined : {
|
|
749
|
-
message: truncate(stderr.trim() || stdout.trim() || result.error?.message || `Cawdex exited with code ${exitCode}`, 2000),
|
|
750
|
-
},
|
|
751
|
-
};
|
|
752
|
-
|
|
753
|
-
emit(output);
|
|
1
|
+
#!/usr/bin/env node
|
|
2
|
+
import { existsSync, mkdirSync, readFileSync, readdirSync, statSync, writeFileSync } from 'node:fs';
|
|
3
|
+
import { tmpdir } from 'node:os';
|
|
4
|
+
import { dirname, join } from 'node:path';
|
|
5
|
+
import { spawnSync } from 'node:child_process';
|
|
6
|
+
|
|
7
|
+
const startedAt = Date.now();
|
|
8
|
+
|
|
9
|
+
function emit(output) {
|
|
10
|
+
process.stdout.write(JSON.stringify(output));
|
|
11
|
+
}
|
|
12
|
+
|
|
13
|
+
function fail(status, message, extra = {}) {
|
|
14
|
+
emit({
|
|
15
|
+
ok: false,
|
|
16
|
+
status,
|
|
17
|
+
failureKind: status,
|
|
18
|
+
finalText: message,
|
|
19
|
+
elapsedMs: Date.now() - startedAt,
|
|
20
|
+
artifacts: [],
|
|
21
|
+
error: { message },
|
|
22
|
+
...extra,
|
|
23
|
+
});
|
|
24
|
+
}
|
|
25
|
+
|
|
26
|
+
function readInput() {
|
|
27
|
+
const inputPath = process.env.KBENCH_ADAPTER_INPUT;
|
|
28
|
+
const raw = inputPath ? readFileSync(inputPath, 'utf8') : readFileSync(0, 'utf8');
|
|
29
|
+
return JSON.parse(raw);
|
|
30
|
+
}
|
|
31
|
+
|
|
32
|
+
function redact(text) {
|
|
33
|
+
return String(text || '')
|
|
34
|
+
.replace(/sk-or-v1-[A-Za-z0-9_-]+/g, 'sk-or-v1-[REDACTED]')
|
|
35
|
+
.replace(/sk-[A-Za-z0-9_-]{16,}/g, 'sk-[REDACTED]')
|
|
36
|
+
.replace(/hf_[A-Za-z0-9]{16,}/g, 'hf_[REDACTED]')
|
|
37
|
+
.replace(/KGAT_[A-Za-z0-9]{16,}/g, 'KGAT_[REDACTED]')
|
|
38
|
+
.replace(/npm_[A-Za-z0-9]{16,}/g, 'npm_[REDACTED]');
|
|
39
|
+
}
|
|
40
|
+
|
|
41
|
+
function truncate(text, max) {
|
|
42
|
+
const safe = redact(text);
|
|
43
|
+
if (safe.length <= max) return safe;
|
|
44
|
+
return `${safe.slice(0, max - 80)}\n...[truncated ${safe.length - (max - 80)} chars]`;
|
|
45
|
+
}
|
|
46
|
+
|
|
47
|
+
function splitCommand(command) {
|
|
48
|
+
const parts = [];
|
|
49
|
+
let cur = '';
|
|
50
|
+
let quote = null;
|
|
51
|
+
let escaped = false;
|
|
52
|
+
for (const ch of command.trim()) {
|
|
53
|
+
if (escaped) {
|
|
54
|
+
cur += ch;
|
|
55
|
+
escaped = false;
|
|
56
|
+
continue;
|
|
57
|
+
}
|
|
58
|
+
if (ch === '\\' && quote) {
|
|
59
|
+
escaped = true;
|
|
60
|
+
continue;
|
|
61
|
+
}
|
|
62
|
+
if (quote) {
|
|
63
|
+
if (ch === quote) quote = null;
|
|
64
|
+
else cur += ch;
|
|
65
|
+
continue;
|
|
66
|
+
}
|
|
67
|
+
if (ch === '"' || ch === "'") {
|
|
68
|
+
quote = ch;
|
|
69
|
+
continue;
|
|
70
|
+
}
|
|
71
|
+
if (/\s/.test(ch)) {
|
|
72
|
+
if (cur) {
|
|
73
|
+
parts.push(cur);
|
|
74
|
+
cur = '';
|
|
75
|
+
}
|
|
76
|
+
continue;
|
|
77
|
+
}
|
|
78
|
+
cur += ch;
|
|
79
|
+
}
|
|
80
|
+
if (cur) parts.push(cur);
|
|
81
|
+
return parts;
|
|
82
|
+
}
|
|
83
|
+
|
|
84
|
+
function profileForBenchmark(benchmark) {
|
|
85
|
+
const slug = String(benchmark || '').toLowerCase().replace(/[^a-z0-9]+/g, '');
|
|
86
|
+
if (slug === 'swe' || slug === 'swebench') return 'swe-bench';
|
|
87
|
+
if (slug === 'tb2' || slug === 'terminalbench') return 'terminal-bench';
|
|
88
|
+
if (slug === 'terminalworld' || slug === 'terminalworldbench' || slug === 'tw' || slug === 'tworld') return 'terminalworld';
|
|
89
|
+
if (slug === 'swechain' || slug === 'chain' || slug === 'upgrade') return 'swe-chain';
|
|
90
|
+
if (slug === 'swecycle' || slug === 'swecyclebench' || slug === 'fullcycle' || slug === 'swejudge') return 'swe-cycle';
|
|
91
|
+
if (slug === 'sweci' || slug === 'swecibench') return 'swe-ci';
|
|
92
|
+
if (slug === 'swepr' || slug === 'sweprbench' || slug === 'prbench' || slug === 'prreview' || slug === 'pullrequestreview' || slug === 'codereviewbench') return 'swe-prbench';
|
|
93
|
+
if (slug === 'tml' || slug === 'tmlbench' || slug === 'tabularml' || slug === 'kaggleml' || slug === 'kagglebench' || slug === 'datascience') return 'tml-bench';
|
|
94
|
+
if (slug === 'pi' || slug === 'pibench' || slug === 'proactive' || slug === 'proactiveassistant' || slug === 'personalassistant' || slug === 'hiddenintent') return 'pi-bench';
|
|
95
|
+
if (slug === 'cirepair' || slug === 'cirepairbench' || slug === 'ci') return 'ci-repair';
|
|
96
|
+
if (slug === 'wildclaw' || slug === 'wildclawbench' || slug === 'wcbench') return 'wildclaw';
|
|
97
|
+
if (slug === 'arc' || slug === 'arcagi' || slug === 'arcagi3' || slug === 'arcprize') return 'arc-agi';
|
|
98
|
+
if (slug === 'spec' || slug === 'specbench' || slug === 'speccompliance') return 'specbench';
|
|
99
|
+
if (slug === 'rhb' || slug === 'rewardhack' || slug === 'rewardhacking' || slug === 'rewardhackingagents') return 'reward-hacking';
|
|
100
|
+
if (slug === 'roadmap' || slug === 'roadmapbench' || slug === 'longhorizon' || slug === 'versionupgrade') return 'roadmapbench';
|
|
101
|
+
if (slug === 'saas' || slug === 'saasbench' || slug === 'enterprise') return 'saasbench';
|
|
102
|
+
if (slug === 'mobile' || slug === 'swebenchmobile' || slug === 'swemobile' || slug === 'ios') return 'swe-bench-mobile';
|
|
103
|
+
if (slug === 'webdev' || slug === 'webdevbench' || slug === 'swewebdev' || slug === 'swewebdevbench' || slug === 'vibecoding') return 'webdevbench';
|
|
104
|
+
if (slug === 'app' || slug === 'appworld' || slug === 'appworldbench') return 'appworld';
|
|
105
|
+
if (slug === 'browsecomp' || slug === 'browsecompplus' || slug === 'deepresearch' || slug === 'webresearch') return 'browsecomp';
|
|
106
|
+
if (slug === 'tau' || slug === 'tau2' || slug === 'taubench' || slug === 'taubench2' || slug.startsWith('tau2') || slug.startsWith('taubench')) return 'tau2';
|
|
107
|
+
return 'generic';
|
|
108
|
+
}
|
|
109
|
+
|
|
110
|
+
function collectTraceRefs(traceDir) {
|
|
111
|
+
const refs = [];
|
|
112
|
+
if (!traceDir || !existsSync(traceDir)) return refs;
|
|
113
|
+
const stack = [traceDir];
|
|
114
|
+
while (stack.length) {
|
|
115
|
+
const dir = stack.pop();
|
|
116
|
+
for (const entry of readdirSync(dir, { withFileTypes: true })) {
|
|
117
|
+
const full = join(dir, entry.name);
|
|
118
|
+
if (entry.isDirectory()) {
|
|
119
|
+
stack.push(full);
|
|
120
|
+
} else if (entry.name === 'summary.json' || entry.name === 'trace.jsonl') {
|
|
121
|
+
refs.push({
|
|
122
|
+
kind: entry.name === 'trace.jsonl' ? 'cawdex-tool-trace' : 'cawdex-summary',
|
|
123
|
+
path: full,
|
|
124
|
+
contentType: entry.name.endsWith('.jsonl') ? 'application/jsonl' : 'application/json',
|
|
125
|
+
description: `Cawdex ${entry.name}`,
|
|
126
|
+
});
|
|
127
|
+
}
|
|
128
|
+
}
|
|
129
|
+
}
|
|
130
|
+
return refs;
|
|
131
|
+
}
|
|
132
|
+
|
|
133
|
+
function readLatestTraceSummary(traceDir) {
|
|
134
|
+
if (!traceDir || !existsSync(traceDir)) return null;
|
|
135
|
+
const summaries = [];
|
|
136
|
+
const stack = [traceDir];
|
|
137
|
+
while (stack.length) {
|
|
138
|
+
const dir = stack.pop();
|
|
139
|
+
for (const entry of readdirSync(dir, { withFileTypes: true })) {
|
|
140
|
+
const full = join(dir, entry.name);
|
|
141
|
+
if (entry.isDirectory()) {
|
|
142
|
+
stack.push(full);
|
|
143
|
+
} else if (entry.name === 'summary.json') {
|
|
144
|
+
try {
|
|
145
|
+
const summary = JSON.parse(readFileSync(full, 'utf8'));
|
|
146
|
+
const endedAtMs = Number.isFinite(Date.parse(summary.endedAt)) ? Date.parse(summary.endedAt) : 0;
|
|
147
|
+
const mtimeMs = statSync(full).mtimeMs;
|
|
148
|
+
summaries.push({ path: full, summary, sortKey: endedAtMs || mtimeMs });
|
|
149
|
+
} catch {
|
|
150
|
+
// Ignore malformed or partially-written trace summaries.
|
|
151
|
+
}
|
|
152
|
+
}
|
|
153
|
+
}
|
|
154
|
+
}
|
|
155
|
+
summaries.sort((a, b) => b.sortKey - a.sortKey);
|
|
156
|
+
return summaries[0] || null;
|
|
157
|
+
}
|
|
158
|
+
|
|
159
|
+
function compactTraceSummary(traceSummary) {
|
|
160
|
+
if (!traceSummary) return undefined;
|
|
161
|
+
const summary = traceSummary.summary || {};
|
|
162
|
+
const quality = summary.trajectoryQuality || {};
|
|
163
|
+
const experienceCard = compactExperienceCard(summary.experienceCard);
|
|
164
|
+
return {
|
|
165
|
+
path: traceSummary.path,
|
|
166
|
+
verificationCount: summary.verificationCount,
|
|
167
|
+
verificationCommands: Array.isArray(summary.verificationCommands) ? summary.verificationCommands.slice(0, 20) : [],
|
|
168
|
+
verificationEvidence: summary.verificationEvidence,
|
|
169
|
+
finalAnswerEvidence: summary.finalAnswerEvidence,
|
|
170
|
+
usage: summary.usage,
|
|
171
|
+
experienceCard,
|
|
172
|
+
agentContextCompilation: compactAgentContextCompilation(summary.agentContextCompilation),
|
|
173
|
+
changeEvaluation: compactChangeEvaluation(summary.changeEvaluation),
|
|
174
|
+
submissionBundleManifest: compactSubmissionBundleManifest(summary.submissionBundleManifest),
|
|
175
|
+
changedFiles: Array.isArray(summary.changedFiles) ? summary.changedFiles.slice(0, 100) : [],
|
|
176
|
+
worktreeChangedFiles: Array.isArray(summary.worktreeChangedFiles) ? summary.worktreeChangedFiles.slice(0, 100) : [],
|
|
177
|
+
artifacts: Array.isArray(summary.artifacts) ? summary.artifacts.slice(0, 20) : [],
|
|
178
|
+
trajectoryQuality: {
|
|
179
|
+
benchmarkContextUsed: quality.benchmarkContextUsed,
|
|
180
|
+
usageCallCount: quality.usageCallCount,
|
|
181
|
+
usageTotalTokens: quality.usageTotalTokens,
|
|
182
|
+
usageEstimatedCostUsd: quality.usageEstimatedCostUsd,
|
|
183
|
+
costEfficiencyRisk: quality.costEfficiencyRisk,
|
|
184
|
+
totalToolElapsedMs: quality.totalToolElapsedMs,
|
|
185
|
+
maxToolElapsedMs: quality.maxToolElapsedMs,
|
|
186
|
+
slowToolCallCount: quality.slowToolCallCount,
|
|
187
|
+
slowToolEvents: Array.isArray(quality.slowToolEvents) ? quality.slowToolEvents.slice(0, 20) : [],
|
|
188
|
+
timeEfficiencyRisk: quality.timeEfficiencyRisk,
|
|
189
|
+
invalidToolActionCount: quality.invalidToolActionCount,
|
|
190
|
+
invalidToolActionPercent: quality.invalidToolActionPercent,
|
|
191
|
+
invalidToolActionEvents: Array.isArray(quality.invalidToolActionEvents) ? quality.invalidToolActionEvents.slice(0, 20) : [],
|
|
192
|
+
localizationBeforeFirstEdit: quality.localizationBeforeFirstEdit,
|
|
193
|
+
failingReproductionBeforeFirstEdit: quality.failingReproductionBeforeFirstEdit,
|
|
194
|
+
passingValidationAfterFirstEdit: quality.passingValidationAfterFirstEdit,
|
|
195
|
+
validationAfterLastEdit: quality.validationAfterLastEdit,
|
|
196
|
+
passingValidationAfterLastEdit: quality.passingValidationAfterLastEdit,
|
|
197
|
+
finalEditVerificationCount: quality.finalEditVerificationCount,
|
|
198
|
+
finalEditPassingVerificationCount: quality.finalEditPassingVerificationCount,
|
|
199
|
+
stableValidationAfterLastEdit: quality.stableValidationAfterLastEdit,
|
|
200
|
+
broadValidationAfterLastEdit: quality.broadValidationAfterLastEdit,
|
|
201
|
+
passingBroadValidationAfterLastEdit: quality.passingBroadValidationAfterLastEdit,
|
|
202
|
+
successfulVerificationCount: quality.successfulVerificationCount,
|
|
203
|
+
failedVerificationCount: quality.failedVerificationCount,
|
|
204
|
+
incompleteVerifierCount: quality.incompleteVerifierCount,
|
|
205
|
+
incompleteVerifierEvents: Array.isArray(quality.incompleteVerifierEvents) ? quality.incompleteVerifierEvents.slice(0, 20) : [],
|
|
206
|
+
inconclusiveVerifierEvents: Array.isArray(quality.inconclusiveVerifierEvents) ? quality.inconclusiveVerifierEvents.slice(0, 20) : [],
|
|
207
|
+
environmentSetupFailureCount: quality.environmentSetupFailureCount,
|
|
208
|
+
environmentSetupFailureEvents: Array.isArray(quality.environmentSetupFailureEvents) ? quality.environmentSetupFailureEvents.slice(0, 20) : [],
|
|
209
|
+
unresolvedEnvironmentSetupFailureCount: quality.unresolvedEnvironmentSetupFailureCount,
|
|
210
|
+
unresolvedEnvironmentSetupFailureEvents: Array.isArray(quality.unresolvedEnvironmentSetupFailureEvents) ? quality.unresolvedEnvironmentSetupFailureEvents.slice(0, 20) : [],
|
|
211
|
+
environmentSetupCount: quality.environmentSetupCount,
|
|
212
|
+
successfulEnvironmentSetupCount: quality.successfulEnvironmentSetupCount,
|
|
213
|
+
environmentSetupEvents: Array.isArray(quality.environmentSetupEvents) ? quality.environmentSetupEvents.slice(0, 20) : [],
|
|
214
|
+
dependencyManifestEditCount: quality.dependencyManifestEditCount,
|
|
215
|
+
dependencyLockfileEditCount: quality.dependencyLockfileEditCount,
|
|
216
|
+
dependencyManifestEditEvents: Array.isArray(quality.dependencyManifestEditEvents) ? quality.dependencyManifestEditEvents.slice(0, 20) : [],
|
|
217
|
+
dependencyLockfileEditEvents: Array.isArray(quality.dependencyLockfileEditEvents) ? quality.dependencyLockfileEditEvents.slice(0, 20) : [],
|
|
218
|
+
dependencySetupAfterManifestEdit: quality.dependencySetupAfterManifestEdit,
|
|
219
|
+
passingDependencySetupAfterManifestEdit: quality.passingDependencySetupAfterManifestEdit,
|
|
220
|
+
dependencyValidationAfterManifestEdit: quality.dependencyValidationAfterManifestEdit,
|
|
221
|
+
passingDependencyValidationAfterManifestEdit: quality.passingDependencyValidationAfterManifestEdit,
|
|
222
|
+
firstDependencySetupAfterManifestEditSeq: quality.firstDependencySetupAfterManifestEditSeq,
|
|
223
|
+
firstDependencyValidationAfterManifestEditSeq: quality.firstDependencyValidationAfterManifestEditSeq,
|
|
224
|
+
skillViewCount: quality.skillViewCount,
|
|
225
|
+
skillViewEvents: Array.isArray(quality.skillViewEvents) ? quality.skillViewEvents.slice(0, 20) : [],
|
|
226
|
+
skillNames: Array.isArray(quality.skillNames) ? quality.skillNames.slice(0, 20) : [],
|
|
227
|
+
skillLoadedBeforeLocalContext: quality.skillLoadedBeforeLocalContext,
|
|
228
|
+
excessiveSkillViewCount: quality.excessiveSkillViewCount,
|
|
229
|
+
ciWorkflowCommandCount: quality.ciWorkflowCommandCount,
|
|
230
|
+
ciVerifierCommands: Array.isArray(quality.ciVerifierCommands) ? quality.ciVerifierCommands.slice(0, 20) : [],
|
|
231
|
+
ciValidationAfterFirstEdit: quality.ciValidationAfterFirstEdit,
|
|
232
|
+
passingCiValidationAfterFirstEdit: quality.passingCiValidationAfterFirstEdit,
|
|
233
|
+
ciValidationAfterLastEdit: quality.ciValidationAfterLastEdit,
|
|
234
|
+
passingCiValidationAfterLastEdit: quality.passingCiValidationAfterLastEdit,
|
|
235
|
+
firstCiValidationAfterFirstEditSeq: quality.firstCiValidationAfterFirstEditSeq,
|
|
236
|
+
sourceResearchCoverage: quality.sourceResearchCoverage,
|
|
237
|
+
taskContractSignalCount: quality.taskContractSignalCount,
|
|
238
|
+
taskContractChecklistUsed: quality.taskContractChecklistUsed,
|
|
239
|
+
taskContractChecklistAfterContext: quality.taskContractChecklistAfterContext,
|
|
240
|
+
taskContractChecklistComplete: quality.taskContractChecklistComplete,
|
|
241
|
+
latestTodoSeq: quality.latestTodoSeq,
|
|
242
|
+
todoIncompleteCount: quality.todoIncompleteCount,
|
|
243
|
+
todoIncompleteItems: Array.isArray(quality.todoIncompleteItems) ? quality.todoIncompleteItems.slice(0, 20) : [],
|
|
244
|
+
taskAlignmentRisk: quality.taskAlignmentRisk,
|
|
245
|
+
taskAlignmentSignalCount: quality.taskAlignmentSignalCount,
|
|
246
|
+
taskAlignmentSignals: Array.isArray(quality.taskAlignmentSignals) ? quality.taskAlignmentSignals.slice(0, 20) : [],
|
|
247
|
+
specComplianceRisk: quality.specComplianceRisk,
|
|
248
|
+
specComplianceSignalCount: quality.specComplianceSignalCount,
|
|
249
|
+
specComplianceSignals: Array.isArray(quality.specComplianceSignals) ? quality.specComplianceSignals.slice(0, 20) : [],
|
|
250
|
+
rewardHackRisk: quality.rewardHackRisk,
|
|
251
|
+
rewardHackSignalCount: quality.rewardHackSignalCount,
|
|
252
|
+
rewardHackSignals: Array.isArray(quality.rewardHackSignals) ? quality.rewardHackSignals.slice(0, 20) : [],
|
|
253
|
+
longHorizonRisk: quality.longHorizonRisk,
|
|
254
|
+
longHorizonSignalCount: quality.longHorizonSignalCount,
|
|
255
|
+
longHorizonSignals: Array.isArray(quality.longHorizonSignals) ? quality.longHorizonSignals.slice(0, 20) : [],
|
|
256
|
+
proactivityDetected: quality.proactivityDetected,
|
|
257
|
+
proactivityRisk: quality.proactivityRisk,
|
|
258
|
+
proactivitySignalCount: quality.proactivitySignalCount,
|
|
259
|
+
proactivitySignals: Array.isArray(quality.proactivitySignals) ? quality.proactivitySignals.slice(0, 20) : [],
|
|
260
|
+
proactivityContextContract: quality.proactivityContextContract,
|
|
261
|
+
proactivityHiddenIntentEvidence: quality.proactivityHiddenIntentEvidence,
|
|
262
|
+
proactivityClarificationEvidence: quality.proactivityClarificationEvidence,
|
|
263
|
+
proactivityPrivacyEvidence: quality.proactivityPrivacyEvidence,
|
|
264
|
+
proactivityCompletionEvidence: quality.proactivityCompletionEvidence,
|
|
265
|
+
proactivityActionCount: quality.proactivityActionCount,
|
|
266
|
+
noEditContractDetected: quality.noEditContractDetected,
|
|
267
|
+
editAfterNoEditContract: quality.editAfterNoEditContract,
|
|
268
|
+
lastEditSeq: quality.lastEditSeq,
|
|
269
|
+
editTargetCount: quality.editTargetCount,
|
|
270
|
+
localizedEditTargetCount: quality.localizedEditTargetCount,
|
|
271
|
+
unlocalizedEditTargetEvents: Array.isArray(quality.unlocalizedEditTargetEvents) ? quality.unlocalizedEditTargetEvents.slice(0, 20) : [],
|
|
272
|
+
contextUtilizationInspectCount: quality.contextUtilizationInspectCount,
|
|
273
|
+
contextUtilizationHitCount: quality.contextUtilizationHitCount,
|
|
274
|
+
contextUtilizationMissCount: quality.contextUtilizationMissCount,
|
|
275
|
+
contextUtilizationPercent: quality.contextUtilizationPercent,
|
|
276
|
+
contextUtilizationRisk: quality.contextUtilizationRisk,
|
|
277
|
+
contextUtilizationMissEvents: Array.isArray(quality.contextUtilizationMissEvents) ? quality.contextUtilizationMissEvents.slice(0, 20) : [],
|
|
278
|
+
preEditContextInspectCount: quality.preEditContextInspectCount,
|
|
279
|
+
preEditContextHitCount: quality.preEditContextHitCount,
|
|
280
|
+
preEditContextMissCount: quality.preEditContextMissCount,
|
|
281
|
+
preEditContextUtilizationPercent: quality.preEditContextUtilizationPercent,
|
|
282
|
+
contextBloatRisk: quality.contextBloatRisk,
|
|
283
|
+
contextBloatEventCount: quality.contextBloatEventCount,
|
|
284
|
+
contextBloatEvents: Array.isArray(quality.contextBloatEvents) ? quality.contextBloatEvents.slice(0, 20) : [],
|
|
285
|
+
evidenceGroundingRisk: quality.evidenceGroundingRisk,
|
|
286
|
+
evidenceGroundingEventCount: quality.evidenceGroundingEventCount,
|
|
287
|
+
evidenceGroundingEvents: Array.isArray(quality.evidenceGroundingEvents) ? quality.evidenceGroundingEvents.slice(0, 20) : [],
|
|
288
|
+
broadEditContractDetected: quality.broadEditContractDetected,
|
|
289
|
+
largeEditSurfaceTargetCount: quality.largeEditSurfaceTargetCount,
|
|
290
|
+
largeEditSurfaceTargets: Array.isArray(quality.largeEditSurfaceTargets) ? quality.largeEditSurfaceTargets.slice(0, 40) : [],
|
|
291
|
+
redundantToolCallCount: quality.redundantToolCallCount,
|
|
292
|
+
redundantToolCallEvents: Array.isArray(quality.redundantToolCallEvents) ? quality.redundantToolCallEvents.slice(0, 20) : [],
|
|
293
|
+
redundantVerifierCount: quality.redundantVerifierCount,
|
|
294
|
+
redundantVerifierEvents: Array.isArray(quality.redundantVerifierEvents) ? quality.redundantVerifierEvents.slice(0, 20) : [],
|
|
295
|
+
blindRepairCount: quality.blindRepairCount,
|
|
296
|
+
blindRepairEvents: Array.isArray(quality.blindRepairEvents) ? quality.blindRepairEvents.slice(0, 20) : [],
|
|
297
|
+
failureAlignedRepairCount: quality.failureAlignedRepairCount,
|
|
298
|
+
failureUnalignedRepairCount: quality.failureUnalignedRepairCount,
|
|
299
|
+
failureUnalignedRepairEvents: Array.isArray(quality.failureUnalignedRepairEvents) ? quality.failureUnalignedRepairEvents.slice(0, 20) : [],
|
|
300
|
+
postEditRegressionCycleCount: quality.postEditRegressionCycleCount,
|
|
301
|
+
postEditRegressionCycleEvents: Array.isArray(quality.postEditRegressionCycleEvents) ? quality.postEditRegressionCycleEvents.slice(0, 20) : [],
|
|
302
|
+
postSuccessMutationCount: quality.postSuccessMutationCount,
|
|
303
|
+
postSuccessMutationEvents: Array.isArray(quality.postSuccessMutationEvents) ? quality.postSuccessMutationEvents.slice(0, 20) : [],
|
|
304
|
+
predictedEditCount: quality.predictedEditCount,
|
|
305
|
+
unpredictedEditCount: quality.unpredictedEditCount,
|
|
306
|
+
contradictedEditPredictionCount: quality.contradictedEditPredictionCount,
|
|
307
|
+
unverifiedEditPredictionCount: quality.unverifiedEditPredictionCount,
|
|
308
|
+
decisionObservabilityRisk: quality.decisionObservabilityRisk,
|
|
309
|
+
scratchArtifactPermissionDetected: quality.scratchArtifactPermissionDetected,
|
|
310
|
+
scratchArtifactEvents: Array.isArray(quality.scratchArtifactEvents) ? quality.scratchArtifactEvents.slice(0, 20) : [],
|
|
311
|
+
postEditDiffReview: quality.postEditDiffReview,
|
|
312
|
+
diffReviewAfterLastEdit: quality.diffReviewAfterLastEdit,
|
|
313
|
+
firstPostEditDiffReviewSeq: quality.firstPostEditDiffReviewSeq,
|
|
314
|
+
firstDiffReviewAfterLastEditSeq: quality.firstDiffReviewAfterLastEditSeq,
|
|
315
|
+
broadValidationAfterFirstEdit: quality.broadValidationAfterFirstEdit,
|
|
316
|
+
passingBroadValidationAfterFirstEdit: quality.passingBroadValidationAfterFirstEdit,
|
|
317
|
+
firstBroadValidationAfterFirstEditSeq: quality.firstBroadValidationAfterFirstEditSeq,
|
|
318
|
+
lastPostEditVerificationSeq: quality.lastPostEditVerificationSeq,
|
|
319
|
+
lastPostEditVerificationStatus: quality.lastPostEditVerificationStatus,
|
|
320
|
+
lastPostEditVerificationConclusiveFailure: quality.lastPostEditVerificationConclusiveFailure,
|
|
321
|
+
firstConclusiveFailedVerificationSeq: quality.firstConclusiveFailedVerificationSeq,
|
|
322
|
+
testEditPermissionDetected: quality.testEditPermissionDetected,
|
|
323
|
+
testHarnessEditEvents: Array.isArray(quality.testHarnessEditEvents) ? quality.testHarnessEditEvents.slice(0, 20) : [],
|
|
324
|
+
processScore: quality.processScore,
|
|
325
|
+
processDefects: Array.isArray(quality.processDefects) ? quality.processDefects.slice(0, 20) : [],
|
|
326
|
+
warnings: Array.isArray(quality.warnings) ? quality.warnings.slice(0, 20) : [],
|
|
327
|
+
},
|
|
328
|
+
};
|
|
329
|
+
}
|
|
330
|
+
|
|
331
|
+
function compactAgentContextCompilation(compilation) {
|
|
332
|
+
if (!compilation || typeof compilation !== 'object' || Array.isArray(compilation)) return undefined;
|
|
333
|
+
const metadata = compilation.metadata && typeof compilation.metadata === 'object' && !Array.isArray(compilation.metadata)
|
|
334
|
+
? compilation.metadata
|
|
335
|
+
: {};
|
|
336
|
+
return {
|
|
337
|
+
version: compilation.version,
|
|
338
|
+
format: compilation.format,
|
|
339
|
+
task: truncate(compilation.task || '', 2000),
|
|
340
|
+
context: truncate(compilation.context || '', 5000),
|
|
341
|
+
answer: truncate(compilation.answer || '', 2500),
|
|
342
|
+
metadata: {
|
|
343
|
+
sessionId: metadata.sessionId,
|
|
344
|
+
mode: metadata.mode,
|
|
345
|
+
provider: metadata.provider,
|
|
346
|
+
model: metadata.model,
|
|
347
|
+
eventCount: metadata.eventCount,
|
|
348
|
+
contextEventCount: metadata.contextEventCount,
|
|
349
|
+
verificationStatus: metadata.verificationStatus,
|
|
350
|
+
successfulVerificationCount: metadata.successfulVerificationCount,
|
|
351
|
+
processScore: metadata.processScore,
|
|
352
|
+
usageTotalTokens: metadata.usageTotalTokens,
|
|
353
|
+
estimatedCostUsd: metadata.estimatedCostUsd,
|
|
354
|
+
changedFiles: Array.isArray(metadata.changedFiles) ? metadata.changedFiles.slice(0, 100) : [],
|
|
355
|
+
verificationCommands: Array.isArray(metadata.verificationCommands) ? metadata.verificationCommands.slice(0, 20) : [],
|
|
356
|
+
sourceResearchCoverage: metadata.sourceResearchCoverage,
|
|
357
|
+
warnings: Array.isArray(metadata.warnings) ? metadata.warnings.slice(0, 20) : [],
|
|
358
|
+
},
|
|
359
|
+
};
|
|
360
|
+
}
|
|
361
|
+
|
|
362
|
+
function compactChangeEvaluation(changeEvaluation) {
|
|
363
|
+
if (!changeEvaluation || typeof changeEvaluation !== 'object' || Array.isArray(changeEvaluation)) return undefined;
|
|
364
|
+
return {
|
|
365
|
+
version: changeEvaluation.version,
|
|
366
|
+
format: changeEvaluation.format,
|
|
367
|
+
status: changeEvaluation.status,
|
|
368
|
+
accepted: changeEvaluation.accepted,
|
|
369
|
+
reason: truncate(changeEvaluation.reason || '', 500),
|
|
370
|
+
editCount: changeEvaluation.editCount,
|
|
371
|
+
predictedEditCount: changeEvaluation.predictedEditCount,
|
|
372
|
+
unpredictedEditCount: changeEvaluation.unpredictedEditCount,
|
|
373
|
+
confirmedPredictionCount: changeEvaluation.confirmedPredictionCount,
|
|
374
|
+
contradictedPredictionCount: changeEvaluation.contradictedPredictionCount,
|
|
375
|
+
unverifiedPredictionCount: changeEvaluation.unverifiedPredictionCount,
|
|
376
|
+
decisionCoveragePercent: changeEvaluation.decisionCoveragePercent,
|
|
377
|
+
regressionCycleCount: changeEvaluation.regressionCycleCount,
|
|
378
|
+
broadRegressionFailureCount: changeEvaluation.broadRegressionFailureCount,
|
|
379
|
+
predictions: Array.isArray(changeEvaluation.predictions) ? changeEvaluation.predictions.slice(0, 20) : [],
|
|
380
|
+
unpredictedEdits: Array.isArray(changeEvaluation.unpredictedEdits) ? changeEvaluation.unpredictedEdits.slice(0, 20) : [],
|
|
381
|
+
regressionCycles: Array.isArray(changeEvaluation.regressionCycles) ? changeEvaluation.regressionCycles.slice(0, 20) : [],
|
|
382
|
+
recommendedAction: truncate(changeEvaluation.recommendedAction || '', 500),
|
|
383
|
+
};
|
|
384
|
+
}
|
|
385
|
+
|
|
386
|
+
function compactSubmissionBundleManifest(manifest) {
|
|
387
|
+
if (!manifest || typeof manifest !== 'object' || Array.isArray(manifest)) return undefined;
|
|
388
|
+
return {
|
|
389
|
+
version: manifest.version,
|
|
390
|
+
format: manifest.format,
|
|
391
|
+
submissionReady: manifest.submissionReady,
|
|
392
|
+
reason: truncate(manifest.reason || '', 500),
|
|
393
|
+
officialResultRequired: manifest.officialResultRequired,
|
|
394
|
+
missingOfficialFields: Array.isArray(manifest.missingOfficialFields) ? manifest.missingOfficialFields.slice(0, 20) : [],
|
|
395
|
+
benchmark: manifest.benchmark,
|
|
396
|
+
benchmarkName: manifest.benchmarkName,
|
|
397
|
+
sessionId: manifest.sessionId,
|
|
398
|
+
provider: manifest.provider,
|
|
399
|
+
model: manifest.model,
|
|
400
|
+
summaryContainer: manifest.summaryContainer,
|
|
401
|
+
artifacts: Array.isArray(manifest.artifacts) ? manifest.artifacts.slice(0, 50) : [],
|
|
402
|
+
verification: manifest.verification,
|
|
403
|
+
usage: manifest.usage,
|
|
404
|
+
process: manifest.process,
|
|
405
|
+
leaderboardDraft: manifest.leaderboardDraft,
|
|
406
|
+
};
|
|
407
|
+
}
|
|
408
|
+
|
|
409
|
+
function compactExperienceCard(card) {
|
|
410
|
+
if (!card || typeof card !== 'object' || Array.isArray(card)) return undefined;
|
|
411
|
+
return {
|
|
412
|
+
version: card.version,
|
|
413
|
+
replayCheckpoints: Array.isArray(card.replayCheckpoints) ? card.replayCheckpoints.slice(0, 20) : [],
|
|
414
|
+
failureSignatures: Array.isArray(card.failureSignatures) ? card.failureSignatures.slice(0, 10) : [],
|
|
415
|
+
sourceResearchCoverage: card.sourceResearchCoverage,
|
|
416
|
+
taskContract: card.taskContract,
|
|
417
|
+
taskAlignment: compactRiskSignalBlock(card.taskAlignment),
|
|
418
|
+
specCompliance: compactRiskSignalBlock(card.specCompliance),
|
|
419
|
+
rewardHack: compactRiskSignalBlock(card.rewardHack),
|
|
420
|
+
longHorizon: compactRiskSignalBlock(card.longHorizon),
|
|
421
|
+
proactivity: compactProactivity(card.proactivity),
|
|
422
|
+
environmentReconstruction: compactEnvironmentReconstruction(card.environmentReconstruction),
|
|
423
|
+
dependencyUpgrade: compactDependencyUpgrade(card.dependencyUpgrade),
|
|
424
|
+
decisionObservability: compactDecisionObservability(card.decisionObservability),
|
|
425
|
+
validationReliability: compactValidationReliability(card.validationReliability),
|
|
426
|
+
contextUtilization: compactContextUtilization(card.contextUtilization),
|
|
427
|
+
runEfficiency: compactRunEfficiency(card.runEfficiency),
|
|
428
|
+
verificationCommands: Array.isArray(card.verificationCommands) ? card.verificationCommands.slice(0, 20) : [],
|
|
429
|
+
changedFiles: Array.isArray(card.changedFiles) ? card.changedFiles.slice(0, 100) : [],
|
|
430
|
+
warnings: Array.isArray(card.warnings) ? card.warnings.slice(0, 20) : [],
|
|
431
|
+
};
|
|
432
|
+
}
|
|
433
|
+
|
|
434
|
+
function compactRiskSignalBlock(block) {
|
|
435
|
+
if (!block || typeof block !== 'object' || Array.isArray(block)) return undefined;
|
|
436
|
+
return {
|
|
437
|
+
risk: block.risk,
|
|
438
|
+
signalCount: block.signalCount,
|
|
439
|
+
signals: Array.isArray(block.signals) ? block.signals.slice(0, 20) : [],
|
|
440
|
+
};
|
|
441
|
+
}
|
|
442
|
+
|
|
443
|
+
function compactProactivity(proactivity) {
|
|
444
|
+
if (!proactivity || typeof proactivity !== 'object' || Array.isArray(proactivity)) return undefined;
|
|
445
|
+
return {
|
|
446
|
+
detected: proactivity.detected,
|
|
447
|
+
risk: proactivity.risk,
|
|
448
|
+
signalCount: proactivity.signalCount,
|
|
449
|
+
signals: Array.isArray(proactivity.signals) ? proactivity.signals.slice(0, 20) : [],
|
|
450
|
+
contextContract: proactivity.contextContract,
|
|
451
|
+
hiddenIntentEvidence: proactivity.hiddenIntentEvidence,
|
|
452
|
+
clarificationEvidence: proactivity.clarificationEvidence,
|
|
453
|
+
privacyEvidence: proactivity.privacyEvidence,
|
|
454
|
+
completionEvidence: proactivity.completionEvidence,
|
|
455
|
+
actionCount: proactivity.actionCount,
|
|
456
|
+
};
|
|
457
|
+
}
|
|
458
|
+
|
|
459
|
+
function compactRunEfficiency(runEfficiency) {
|
|
460
|
+
if (!runEfficiency || typeof runEfficiency !== 'object' || Array.isArray(runEfficiency)) return undefined;
|
|
461
|
+
return {
|
|
462
|
+
toolCallCount: runEfficiency.toolCallCount,
|
|
463
|
+
totalToolElapsedMs: runEfficiency.totalToolElapsedMs,
|
|
464
|
+
maxToolElapsedMs: runEfficiency.maxToolElapsedMs,
|
|
465
|
+
slowToolCallCount: runEfficiency.slowToolCallCount,
|
|
466
|
+
usageCallCount: runEfficiency.usageCallCount,
|
|
467
|
+
totalTokens: runEfficiency.totalTokens,
|
|
468
|
+
estimatedCostUsd: runEfficiency.estimatedCostUsd,
|
|
469
|
+
successfulVerificationCount: runEfficiency.successfulVerificationCount,
|
|
470
|
+
processScore: runEfficiency.processScore,
|
|
471
|
+
processDefectCount: runEfficiency.processDefectCount,
|
|
472
|
+
warningCount: runEfficiency.warningCount,
|
|
473
|
+
invalidToolActionCount: runEfficiency.invalidToolActionCount,
|
|
474
|
+
invalidToolActionPercent: runEfficiency.invalidToolActionPercent,
|
|
475
|
+
costEfficiencyRisk: runEfficiency.costEfficiencyRisk,
|
|
476
|
+
timeEfficiencyRisk: runEfficiency.timeEfficiencyRisk,
|
|
477
|
+
slowToolEvents: Array.isArray(runEfficiency.slowToolEvents) ? runEfficiency.slowToolEvents.slice(0, 12) : [],
|
|
478
|
+
};
|
|
479
|
+
}
|
|
480
|
+
|
|
481
|
+
function compactContextUtilization(contextUtilization) {
|
|
482
|
+
if (!contextUtilization || typeof contextUtilization !== 'object' || Array.isArray(contextUtilization)) return undefined;
|
|
483
|
+
return {
|
|
484
|
+
inspectCount: contextUtilization.inspectCount,
|
|
485
|
+
hitCount: contextUtilization.hitCount,
|
|
486
|
+
missCount: contextUtilization.missCount,
|
|
487
|
+
utilizationPercent: contextUtilization.utilizationPercent,
|
|
488
|
+
risk: contextUtilization.risk,
|
|
489
|
+
missEvents: Array.isArray(contextUtilization.missEvents) ? contextUtilization.missEvents.slice(0, 12) : [],
|
|
490
|
+
};
|
|
491
|
+
}
|
|
492
|
+
|
|
493
|
+
function compactValidationReliability(validationReliability) {
|
|
494
|
+
if (!validationReliability || typeof validationReliability !== 'object' || Array.isArray(validationReliability)) return undefined;
|
|
495
|
+
return {
|
|
496
|
+
lastEditSeq: validationReliability.lastEditSeq,
|
|
497
|
+
finalEditVerificationCount: validationReliability.finalEditVerificationCount,
|
|
498
|
+
finalEditPassingVerificationCount: validationReliability.finalEditPassingVerificationCount,
|
|
499
|
+
stableValidationAfterLastEdit: validationReliability.stableValidationAfterLastEdit,
|
|
500
|
+
broadValidationAfterLastEdit: validationReliability.broadValidationAfterLastEdit,
|
|
501
|
+
passingBroadValidationAfterLastEdit: validationReliability.passingBroadValidationAfterLastEdit,
|
|
502
|
+
ciValidationAfterLastEdit: validationReliability.ciValidationAfterLastEdit,
|
|
503
|
+
passingCiValidationAfterLastEdit: validationReliability.passingCiValidationAfterLastEdit,
|
|
504
|
+
postEditRegressionCycleCount: validationReliability.postEditRegressionCycleCount,
|
|
505
|
+
lastPostEditVerificationSeq: validationReliability.lastPostEditVerificationSeq,
|
|
506
|
+
lastPostEditVerificationStatus: validationReliability.lastPostEditVerificationStatus,
|
|
507
|
+
finalVerifierCommands: Array.isArray(validationReliability.finalVerifierCommands) ? validationReliability.finalVerifierCommands.slice(0, 12) : [],
|
|
508
|
+
};
|
|
509
|
+
}
|
|
510
|
+
|
|
511
|
+
function compactDecisionObservability(decisionObservability) {
|
|
512
|
+
if (!decisionObservability || typeof decisionObservability !== 'object' || Array.isArray(decisionObservability)) return undefined;
|
|
513
|
+
return {
|
|
514
|
+
editCount: decisionObservability.editCount,
|
|
515
|
+
predictedEditCount: decisionObservability.predictedEditCount,
|
|
516
|
+
verifiedPredictionCount: decisionObservability.verifiedPredictionCount,
|
|
517
|
+
editPredictions: Array.isArray(decisionObservability.editPredictions) ? decisionObservability.editPredictions.slice(0, 12) : [],
|
|
518
|
+
};
|
|
519
|
+
}
|
|
520
|
+
|
|
521
|
+
function compactEnvironmentReconstruction(environmentReconstruction) {
|
|
522
|
+
if (!environmentReconstruction || typeof environmentReconstruction !== 'object' || Array.isArray(environmentReconstruction)) return undefined;
|
|
523
|
+
return {
|
|
524
|
+
setupFailureCount: environmentReconstruction.setupFailureCount,
|
|
525
|
+
unresolvedSetupFailureCount: environmentReconstruction.unresolvedSetupFailureCount,
|
|
526
|
+
setupCount: environmentReconstruction.setupCount,
|
|
527
|
+
successfulSetupCount: environmentReconstruction.successfulSetupCount,
|
|
528
|
+
setupEvents: Array.isArray(environmentReconstruction.setupEvents) ? environmentReconstruction.setupEvents.slice(0, 12) : [],
|
|
529
|
+
setupFailures: Array.isArray(environmentReconstruction.setupFailures) ? environmentReconstruction.setupFailures.slice(0, 12) : [],
|
|
530
|
+
unresolvedSetupFailures: Array.isArray(environmentReconstruction.unresolvedSetupFailures) ? environmentReconstruction.unresolvedSetupFailures.slice(0, 12) : [],
|
|
531
|
+
};
|
|
532
|
+
}
|
|
533
|
+
|
|
534
|
+
function compactDependencyUpgrade(dependencyUpgrade) {
|
|
535
|
+
if (!dependencyUpgrade || typeof dependencyUpgrade !== 'object' || Array.isArray(dependencyUpgrade)) return undefined;
|
|
536
|
+
return {
|
|
537
|
+
manifestEditCount: dependencyUpgrade.manifestEditCount,
|
|
538
|
+
lockfileEditCount: dependencyUpgrade.lockfileEditCount,
|
|
539
|
+
manifestEdits: Array.isArray(dependencyUpgrade.manifestEdits) ? dependencyUpgrade.manifestEdits.slice(0, 12) : [],
|
|
540
|
+
lockfileEdits: Array.isArray(dependencyUpgrade.lockfileEdits) ? dependencyUpgrade.lockfileEdits.slice(0, 12) : [],
|
|
541
|
+
setupAfterManifestEdit: dependencyUpgrade.setupAfterManifestEdit,
|
|
542
|
+
passingSetupAfterManifestEdit: dependencyUpgrade.passingSetupAfterManifestEdit,
|
|
543
|
+
validationAfterManifestEdit: dependencyUpgrade.validationAfterManifestEdit,
|
|
544
|
+
passingValidationAfterManifestEdit: dependencyUpgrade.passingValidationAfterManifestEdit,
|
|
545
|
+
firstSetupAfterManifestEditSeq: dependencyUpgrade.firstSetupAfterManifestEditSeq,
|
|
546
|
+
firstValidationAfterManifestEditSeq: dependencyUpgrade.firstValidationAfterManifestEditSeq,
|
|
547
|
+
};
|
|
548
|
+
}
|
|
549
|
+
|
|
550
|
+
function collectGitArtifactRefs(workdir, artifactRoot) {
|
|
551
|
+
const refs = [];
|
|
552
|
+
if (!workdir || !existsSync(workdir)) return refs;
|
|
553
|
+
|
|
554
|
+
const gitCheck = spawnSync('git', ['-C', workdir, 'rev-parse', '--is-inside-work-tree'], {
|
|
555
|
+
encoding: 'utf8',
|
|
556
|
+
timeout: 5000,
|
|
557
|
+
maxBuffer: 64 * 1024,
|
|
558
|
+
});
|
|
559
|
+
if (gitCheck.status !== 0 || !String(gitCheck.stdout || '').trim().startsWith('true')) {
|
|
560
|
+
return refs;
|
|
561
|
+
}
|
|
562
|
+
|
|
563
|
+
const diff = buildWorktreePatch(workdir);
|
|
564
|
+
if (diff.trim()) {
|
|
565
|
+
const patchPath = join(artifactRoot, 'cawdex.patch');
|
|
566
|
+
writeFileSync(patchPath, redact(diff), 'utf8');
|
|
567
|
+
refs.push({
|
|
568
|
+
kind: 'patch',
|
|
569
|
+
path: patchPath,
|
|
570
|
+
contentType: 'text/x-diff',
|
|
571
|
+
description: 'Redacted git diff after Cawdex run.',
|
|
572
|
+
});
|
|
573
|
+
}
|
|
574
|
+
|
|
575
|
+
const status = spawnSync('git', ['-C', workdir, 'status', '--short'], {
|
|
576
|
+
encoding: 'utf8',
|
|
577
|
+
timeout: 5000,
|
|
578
|
+
maxBuffer: 512 * 1024,
|
|
579
|
+
});
|
|
580
|
+
if (status.stdout && status.stdout.trim()) {
|
|
581
|
+
const statusPath = join(artifactRoot, 'git-status.txt');
|
|
582
|
+
writeFileSync(statusPath, redact(status.stdout), 'utf8');
|
|
583
|
+
refs.push({
|
|
584
|
+
kind: 'git-status',
|
|
585
|
+
path: statusPath,
|
|
586
|
+
contentType: 'text/plain',
|
|
587
|
+
description: 'Redacted git status after Cawdex run.',
|
|
588
|
+
});
|
|
589
|
+
}
|
|
590
|
+
|
|
591
|
+
return refs;
|
|
592
|
+
}
|
|
593
|
+
|
|
594
|
+
function runGit(workdir, args, options = {}) {
|
|
595
|
+
const result = spawnSync('git', ['-C', workdir, ...args], {
|
|
596
|
+
encoding: 'utf8',
|
|
597
|
+
timeout: options.timeout || 5000,
|
|
598
|
+
maxBuffer: options.maxBuffer || 1024 * 1024,
|
|
599
|
+
});
|
|
600
|
+
if (result.error) return '';
|
|
601
|
+
if (result.status !== 0 && !(options.allowDiffExit && result.status === 1)) return '';
|
|
602
|
+
return result.stdout || '';
|
|
603
|
+
}
|
|
604
|
+
|
|
605
|
+
function buildWorktreePatch(workdir) {
|
|
606
|
+
const parts = [
|
|
607
|
+
runGit(workdir, ['diff', '--binary', '--no-ext-diff'], { timeout: 10000, maxBuffer: 20 * 1024 * 1024 }),
|
|
608
|
+
runGit(workdir, ['diff', '--cached', '--binary', '--no-ext-diff'], { timeout: 10000, maxBuffer: 20 * 1024 * 1024 }),
|
|
609
|
+
...collectUntrackedDiffs(workdir),
|
|
610
|
+
].map((part) => part.trim()).filter(Boolean);
|
|
611
|
+
return parts.join('\n\n') + (parts.length ? '\n' : '');
|
|
612
|
+
}
|
|
613
|
+
|
|
614
|
+
function collectUntrackedDiffs(workdir) {
|
|
615
|
+
const raw = runGit(workdir, ['ls-files', '--others', '--exclude-standard', '-z']);
|
|
616
|
+
const files = raw.split('\0').map((file) => file.trim()).filter(Boolean).slice(0, 80);
|
|
617
|
+
return files
|
|
618
|
+
.map((file) => runGit(workdir, ['diff', '--no-index', '--binary', '--no-ext-diff', '--', '/dev/null', file], {
|
|
619
|
+
timeout: 5000,
|
|
620
|
+
maxBuffer: 5 * 1024 * 1024,
|
|
621
|
+
allowDiffExit: true,
|
|
622
|
+
}))
|
|
623
|
+
.filter((diff) => diff.trim());
|
|
624
|
+
}
|
|
625
|
+
|
|
626
|
+
let payload;
|
|
627
|
+
try {
|
|
628
|
+
payload = readInput();
|
|
629
|
+
} catch (err) {
|
|
630
|
+
fail('invalid_adapter', `Could not read adapter input: ${err?.message || err}`);
|
|
631
|
+
process.exit(0);
|
|
632
|
+
}
|
|
633
|
+
|
|
634
|
+
if (payload.mode !== 'task') {
|
|
635
|
+
fail('unsupported_capability', 'Cawdex KBench adapter currently supports task mode only.');
|
|
636
|
+
process.exit(0);
|
|
637
|
+
}
|
|
638
|
+
|
|
639
|
+
const task = payload.task || {};
|
|
640
|
+
const config = payload.config || {};
|
|
641
|
+
const env = payload.env || task.env || {};
|
|
642
|
+
const taskEnv = task.env || {};
|
|
643
|
+
const benchmark = task.benchmark || 'swe';
|
|
644
|
+
const instruction = String(task.instruction || '').trim();
|
|
645
|
+
const profile = profileForBenchmark(benchmark);
|
|
646
|
+
const prompt = `/benchmark ${profile} ${instruction}`;
|
|
647
|
+
const workdir = config.workDir || env.workdir || env.repoPath || taskEnv.workdir || taskEnv.repoPath || process.cwd();
|
|
648
|
+
const artifactRoot = config.storeDir
|
|
649
|
+
|| process.env.CAWDEX_KBENCH_ARTIFACT_DIR
|
|
650
|
+
|| (() => {
|
|
651
|
+
const dir = join(tmpdir(), `cawdex-kbench-${process.pid}-${Date.now()}`);
|
|
652
|
+
mkdirSync(dir, { recursive: true });
|
|
653
|
+
return dir;
|
|
654
|
+
})();
|
|
655
|
+
mkdirSync(artifactRoot, { recursive: true });
|
|
656
|
+
|
|
657
|
+
const stdoutPath = join(artifactRoot, 'cawdex.stdout.txt');
|
|
658
|
+
const stderrPath = join(artifactRoot, 'cawdex.stderr.txt');
|
|
659
|
+
const instructionPath = join(artifactRoot, 'instruction.txt');
|
|
660
|
+
const traceDir = join(artifactRoot, 'cawdex-trace');
|
|
661
|
+
mkdirSync(dirname(stdoutPath), { recursive: true });
|
|
662
|
+
writeFileSync(instructionPath, redact(instruction), 'utf8');
|
|
663
|
+
|
|
664
|
+
const commandParts = splitCommand(process.env.CAWDEX_KBENCH_COMMAND || process.env.CAWDEX_KBENCH_COMMAND || 'cawdex');
|
|
665
|
+
if (!commandParts.length) {
|
|
666
|
+
fail('invalid_adapter', 'CAWDEX_KBENCH_COMMAND resolved to an empty command.');
|
|
667
|
+
process.exit(0);
|
|
668
|
+
}
|
|
669
|
+
|
|
670
|
+
const [command, ...prefixArgs] = commandParts;
|
|
671
|
+
const args = [
|
|
672
|
+
...prefixArgs,
|
|
673
|
+
'--prompt', prompt,
|
|
674
|
+
'--perm', process.env.CAWDEX_KBENCH_PERMISSION || 'yolo',
|
|
675
|
+
'--output-format', 'text',
|
|
676
|
+
'--benchmark-trace-dir', traceDir,
|
|
677
|
+
];
|
|
678
|
+
if (config.modelName) args.push('--model', String(config.modelName));
|
|
679
|
+
if (config.temperature !== undefined) args.push('--temperature', String(config.temperature));
|
|
680
|
+
if (config.baseUrl) args.push('--base-url', String(config.baseUrl));
|
|
681
|
+
if (config.apiKeyEnv) args.push('--api-key-env', String(config.apiKeyEnv));
|
|
682
|
+
if (process.env.CAWDEX_KBENCH_EXTRA_ARGS) {
|
|
683
|
+
args.push(...splitCommand(process.env.CAWDEX_KBENCH_EXTRA_ARGS));
|
|
684
|
+
}
|
|
685
|
+
|
|
686
|
+
const childEnv = {
|
|
687
|
+
...process.env,
|
|
688
|
+
CAWDEX_BENCHMARK_TRACE: '1',
|
|
689
|
+
CAWDEX_BENCHMARK_TRACE_DIR: traceDir,
|
|
690
|
+
CAWDEX_BASH_TIMEOUT_MS: process.env.CAWDEX_BASH_TIMEOUT_MS || '300000',
|
|
691
|
+
};
|
|
692
|
+
for (const [key, value] of Object.entries(env.envVars || {})) {
|
|
693
|
+
if (typeof value === 'string') childEnv[key] = value;
|
|
694
|
+
}
|
|
695
|
+
for (const [key, value] of Object.entries(taskEnv.envVars || {})) {
|
|
696
|
+
if (typeof value === 'string') childEnv[key] = value;
|
|
697
|
+
}
|
|
698
|
+
|
|
699
|
+
const result = spawnSync(command, args, {
|
|
700
|
+
cwd: existsSync(workdir) ? workdir : process.cwd(),
|
|
701
|
+
env: childEnv,
|
|
702
|
+
encoding: 'utf8',
|
|
703
|
+
timeout: typeof config.timeoutMs === 'number' && config.timeoutMs > 0 ? config.timeoutMs : undefined,
|
|
704
|
+
maxBuffer: 20 * 1024 * 1024,
|
|
705
|
+
});
|
|
706
|
+
|
|
707
|
+
const stdout = result.stdout || '';
|
|
708
|
+
const stderr = result.stderr || '';
|
|
709
|
+
writeFileSync(stdoutPath, redact(stdout), 'utf8');
|
|
710
|
+
writeFileSync(stderrPath, redact(stderr), 'utf8');
|
|
711
|
+
const elapsedMs = Date.now() - startedAt;
|
|
712
|
+
const timedOut = result.error?.code === 'ETIMEDOUT' || result.signal === 'SIGTERM';
|
|
713
|
+
const exitCode = typeof result.status === 'number' ? result.status : (timedOut ? 124 : 1);
|
|
714
|
+
const ok = exitCode === 0;
|
|
715
|
+
const traceRefs = collectTraceRefs(traceDir);
|
|
716
|
+
const traceSummary = compactTraceSummary(readLatestTraceSummary(traceDir));
|
|
717
|
+
const workdirUsed = existsSync(workdir) ? workdir : process.cwd();
|
|
718
|
+
const gitRefs = collectGitArtifactRefs(workdirUsed, artifactRoot);
|
|
719
|
+
const stdoutLines = stdout.split(/\r?\n/).map((line) => line.trim()).filter(Boolean);
|
|
720
|
+
const finalText = stdoutLines.at(-1) || stdout.trim() || (ok ? 'Cawdex completed.' : 'Cawdex produced no stdout.');
|
|
721
|
+
const artifacts = [
|
|
722
|
+
{ kind: 'instruction', path: instructionPath, contentType: 'text/plain', description: 'KBench task instruction passed to Cawdex.' },
|
|
723
|
+
{ kind: 'stdout', path: stdoutPath, contentType: 'text/plain', description: 'Cawdex stdout.' },
|
|
724
|
+
{ kind: 'stderr', path: stderrPath, contentType: 'text/plain', description: 'Cawdex stderr.' },
|
|
725
|
+
...gitRefs,
|
|
726
|
+
...traceRefs,
|
|
727
|
+
];
|
|
728
|
+
|
|
729
|
+
const output = {
|
|
730
|
+
ok,
|
|
731
|
+
status: ok ? 'ok' : (timedOut ? 'timeout' : 'agent_error'),
|
|
732
|
+
failureKind: ok ? undefined : (timedOut ? 'timeout' : `exit_${exitCode}`),
|
|
733
|
+
finalText: truncate(finalText, 4000),
|
|
734
|
+
elapsedMs,
|
|
735
|
+
artifacts,
|
|
736
|
+
trace: traceRefs.length ? { native: traceRefs } : undefined,
|
|
737
|
+
benchmarkResult: {
|
|
738
|
+
mode: 'cawdex-kbench',
|
|
739
|
+
benchmark,
|
|
740
|
+
profile,
|
|
741
|
+
exitCode,
|
|
742
|
+
workdir: workdirUsed,
|
|
743
|
+
traceSummary,
|
|
744
|
+
verificationEvidence: traceSummary?.verificationEvidence,
|
|
745
|
+
experienceCard: traceSummary?.experienceCard,
|
|
746
|
+
usage: traceSummary?.usage,
|
|
747
|
+
},
|
|
748
|
+
error: ok ? undefined : {
|
|
749
|
+
message: truncate(stderr.trim() || stdout.trim() || result.error?.message || `Cawdex exited with code ${exitCode}`, 2000),
|
|
750
|
+
},
|
|
751
|
+
};
|
|
752
|
+
|
|
753
|
+
emit(output);
|