hydro-ai-helper 3.0.3 → 3.1.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (71) hide show
  1. package/README.md +8 -0
  2. package/README_en.md +8 -0
  3. package/dist/benchmarks/testdataBenchmark.js +309 -0
  4. package/dist/benchmarks/testdataBenchmark.js.map +1 -0
  5. package/dist/benchmarks/testdataBenchmarkCases.js +204 -0
  6. package/dist/benchmarks/testdataBenchmarkCases.js.map +1 -0
  7. package/dist/benchmarks/testdataBenchmarkCli.js +312 -0
  8. package/dist/benchmarks/testdataBenchmarkCli.js.map +1 -0
  9. package/dist/constants/jailbreakRules.js +55 -39
  10. package/dist/constants/jailbreakRules.js.map +1 -1
  11. package/dist/handlers/adminConfigHandler.js +384 -32
  12. package/dist/handlers/adminConfigHandler.js.map +1 -1
  13. package/dist/handlers/adminHandler.js +10 -2
  14. package/dist/handlers/adminHandler.js.map +1 -1
  15. package/dist/handlers/studentHandler.js +319 -31
  16. package/dist/handlers/studentHandler.js.map +1 -1
  17. package/dist/handlers/teachingSummaryHandler.js +12 -1
  18. package/dist/handlers/teachingSummaryHandler.js.map +1 -1
  19. package/dist/handlers/testdataGenHandler.js +772 -36
  20. package/dist/handlers/testdataGenHandler.js.map +1 -1
  21. package/dist/index.js +47 -0
  22. package/dist/index.js.map +1 -1
  23. package/dist/models/featureStats.js +56 -2
  24. package/dist/models/featureStats.js.map +1 -1
  25. package/dist/models/jailbreakLog.js +481 -10
  26. package/dist/models/jailbreakLog.js.map +1 -1
  27. package/dist/models/testdataGenerationJob.js +280 -0
  28. package/dist/models/testdataGenerationJob.js.map +1 -0
  29. package/dist/services/effectivenessService.js +11 -1
  30. package/dist/services/effectivenessService.js.map +1 -1
  31. package/dist/services/errorReporter.js +12 -2
  32. package/dist/services/errorReporter.js.map +1 -1
  33. package/dist/services/goJudgeSandboxService.js +280 -15
  34. package/dist/services/goJudgeSandboxService.js.map +1 -1
  35. package/dist/services/migrationService.js +57 -15
  36. package/dist/services/migrationService.js.map +1 -1
  37. package/dist/services/openaiClient.js +56 -13
  38. package/dist/services/openaiClient.js.map +1 -1
  39. package/dist/services/outputSafetyService.js +38 -26
  40. package/dist/services/outputSafetyService.js.map +1 -1
  41. package/dist/services/promptService.js +160 -46
  42. package/dist/services/promptService.js.map +1 -1
  43. package/dist/services/safetyPenaltyService.js +84 -0
  44. package/dist/services/safetyPenaltyService.js.map +1 -0
  45. package/dist/services/teachingAnalysisService.js +3 -0
  46. package/dist/services/teachingAnalysisService.js.map +1 -1
  47. package/dist/services/telemetryService.js +10 -1
  48. package/dist/services/telemetryService.js.map +1 -1
  49. package/dist/services/testdataGenService.js +3535 -350
  50. package/dist/services/testdataGenService.js.map +1 -1
  51. package/dist/types/safety.js +3 -0
  52. package/dist/types/safety.js.map +1 -0
  53. package/frontend/admin/ConfigPanel.tsx +67 -160
  54. package/frontend/admin/JailbreakAnalyticsPanel.tsx +250 -0
  55. package/frontend/admin/JailbreakLogsViewer.tsx +605 -24
  56. package/frontend/admin/SafetyGovernancePanel.tsx +622 -0
  57. package/frontend/admin/configTypes.ts +81 -3
  58. package/frontend/components/AIHelperDashboard.tsx +32 -4
  59. package/frontend/components/ScoreboardTabContainer.tsx +13 -2
  60. package/frontend/generated/localeFallback.ts +483 -25
  61. package/frontend/student/AIAssistantPanel.tsx +50 -15
  62. package/frontend/student/hooks/useChatSession.ts +76 -6
  63. package/frontend/student/types.ts +49 -4
  64. package/frontend/teachingSummary/TeachingReviewPanel.tsx +295 -186
  65. package/frontend/teachingSummary/useTeachingReview.ts +11 -4
  66. package/frontend/teachingSummary/useTeachingSummary.ts +2 -0
  67. package/frontend/testdataGen/TestdataGenPanel.tsx +741 -84
  68. package/locales/en.yaml +242 -13
  69. package/locales/zh.yaml +241 -12
  70. package/package.json +3 -1
  71. package/frontend/test.page.ts +0 -91
@@ -0,0 +1,312 @@
1
+ "use strict";
2
+ var __createBinding = (this && this.__createBinding) || (Object.create ? (function(o, m, k, k2) {
3
+ if (k2 === undefined) k2 = k;
4
+ var desc = Object.getOwnPropertyDescriptor(m, k);
5
+ if (!desc || ("get" in desc ? !m.__esModule : desc.writable || desc.configurable)) {
6
+ desc = { enumerable: true, get: function() { return m[k]; } };
7
+ }
8
+ Object.defineProperty(o, k2, desc);
9
+ }) : (function(o, m, k, k2) {
10
+ if (k2 === undefined) k2 = k;
11
+ o[k2] = m[k];
12
+ }));
13
+ var __setModuleDefault = (this && this.__setModuleDefault) || (Object.create ? (function(o, v) {
14
+ Object.defineProperty(o, "default", { enumerable: true, value: v });
15
+ }) : function(o, v) {
16
+ o["default"] = v;
17
+ });
18
+ var __importStar = (this && this.__importStar) || (function () {
19
+ var ownKeys = function(o) {
20
+ ownKeys = Object.getOwnPropertyNames || function (o) {
21
+ var ar = [];
22
+ for (var k in o) if (Object.prototype.hasOwnProperty.call(o, k)) ar[ar.length] = k;
23
+ return ar;
24
+ };
25
+ return ownKeys(o);
26
+ };
27
+ return function (mod) {
28
+ if (mod && mod.__esModule) return mod;
29
+ var result = {};
30
+ if (mod != null) for (var k = ownKeys(mod), i = 0; i < k.length; i++) if (k[i] !== "default") __createBinding(result, mod, k[i]);
31
+ __setModuleDefault(result, mod);
32
+ return result;
33
+ };
34
+ })();
35
+ var __importDefault = (this && this.__importDefault) || function (mod) {
36
+ return (mod && mod.__esModule) ? mod : { "default": mod };
37
+ };
38
+ Object.defineProperty(exports, "__esModule", { value: true });
39
+ exports.parseTestdataBenchmarkCliArgs = parseTestdataBenchmarkCliArgs;
40
+ exports.main = main;
41
+ const child_process_1 = require("child_process");
42
+ const promises_1 = require("fs/promises");
43
+ const path_1 = __importDefault(require("path"));
44
+ const package_json_1 = __importDefault(require("../../package.json"));
45
+ const testdataBenchmark_1 = require("./testdataBenchmark");
46
+ const testdataBenchmarkCases_1 = require("./testdataBenchmarkCases");
47
+ function usage() {
48
+ return `真实模型测试数据难题基准
49
+
50
+ 用法:
51
+ npm run benchmark:testdata -- --list
52
+ npm run benchmark:testdata -- --confirm-cost [--case=id1,id2] [--output=report.json] [--aggregate-output=safe.json] [--compare=old.json]
53
+ npm run benchmark:testdata -- --compare-reports=old.json,new.json
54
+
55
+ 必需环境变量(--list 除外):
56
+ TESTDATA_BENCHMARK_API_BASE OpenAI 兼容 API base,例如 https://api.openai.com/v1
57
+ TESTDATA_BENCHMARK_API_KEY API key
58
+ TESTDATA_BENCHMARK_MODELS 模型链,逗号分隔;也可用 TESTDATA_BENCHMARK_MODEL 指定单模型
59
+ TESTDATA_BENCHMARK_SANDBOX_HOST go-judge 地址,例如 http://127.0.0.1:5050
60
+
61
+ 可选环境变量:
62
+ TESTDATA_BENCHMARK_TIMEOUT_SECONDS 单次模型调用超时,默认 600
63
+ TESTDATA_BENCHMARK_REVISION 报告版本标识;默认读取当前 Git commit
64
+
65
+ 安全说明:必须显式传 --confirm-cost 才会调用模型;题目串行执行,避免挤兑沙箱。`;
66
+ }
67
+ function readValue(argv, index, name) {
68
+ const current = argv[index];
69
+ const prefix = `${name}=`;
70
+ if (current.startsWith(prefix))
71
+ return { value: current.slice(prefix.length), consumed: 0 };
72
+ const next = argv[index + 1];
73
+ if (!next || next.startsWith('--'))
74
+ throw new Error(`${name} 缺少参数`);
75
+ return { value: next, consumed: 1 };
76
+ }
77
+ function parseTestdataBenchmarkCliArgs(argv) {
78
+ const options = {
79
+ list: false,
80
+ help: false,
81
+ confirmCost: false,
82
+ caseIds: [],
83
+ minPassRate: 1,
84
+ };
85
+ for (let i = 0; i < argv.length; i++) {
86
+ const arg = argv[i];
87
+ if (arg === '--list')
88
+ options.list = true;
89
+ else if (arg === '--help' || arg === '-h')
90
+ options.help = true;
91
+ else if (arg === '--confirm-cost')
92
+ options.confirmCost = true;
93
+ else if (arg === '--case' || arg.startsWith('--case=')) {
94
+ const parsed = readValue(argv, i, '--case');
95
+ options.caseIds.push(...parsed.value.split(',').map(value => value.trim()).filter(Boolean));
96
+ i += parsed.consumed;
97
+ }
98
+ else if (arg === '--output' || arg.startsWith('--output=')) {
99
+ const parsed = readValue(argv, i, '--output');
100
+ options.output = parsed.value;
101
+ i += parsed.consumed;
102
+ }
103
+ else if (arg === '--aggregate-output' || arg.startsWith('--aggregate-output=')) {
104
+ const parsed = readValue(argv, i, '--aggregate-output');
105
+ options.aggregateOutput = parsed.value;
106
+ i += parsed.consumed;
107
+ }
108
+ else if (arg === '--compare' || arg.startsWith('--compare=')) {
109
+ const parsed = readValue(argv, i, '--compare');
110
+ options.compare = parsed.value;
111
+ i += parsed.consumed;
112
+ }
113
+ else if (arg === '--compare-reports' || arg.startsWith('--compare-reports=')) {
114
+ const parsed = readValue(argv, i, '--compare-reports');
115
+ const paths = parsed.value.split(',').map(value => value.trim()).filter(Boolean);
116
+ if (paths.length !== 2)
117
+ throw new Error('--compare-reports 必须提供 old.json,new.json');
118
+ options.compareReports = [paths[0], paths[1]];
119
+ i += parsed.consumed;
120
+ }
121
+ else if (arg === '--min-pass-rate' || arg.startsWith('--min-pass-rate=')) {
122
+ const parsed = readValue(argv, i, '--min-pass-rate');
123
+ options.minPassRate = Number(parsed.value);
124
+ i += parsed.consumed;
125
+ }
126
+ else {
127
+ throw new Error(`未知参数:${arg}`);
128
+ }
129
+ }
130
+ if (!Number.isFinite(options.minPassRate)
131
+ || options.minPassRate < 0
132
+ || options.minPassRate > 1) {
133
+ throw new Error('--min-pass-rate 必须是 0 到 1 之间的数字');
134
+ }
135
+ options.caseIds = [...new Set(options.caseIds)];
136
+ return options;
137
+ }
138
+ function requiredEnv(name) {
139
+ const value = process.env[name]?.trim();
140
+ if (!value)
141
+ throw new Error(`缺少环境变量 ${name}`);
142
+ return value;
143
+ }
144
+ function buildModels() {
145
+ const apiBaseUrl = requiredEnv('TESTDATA_BENCHMARK_API_BASE');
146
+ const apiKey = requiredEnv('TESTDATA_BENCHMARK_API_KEY');
147
+ const modelValue = process.env.TESTDATA_BENCHMARK_MODELS?.trim()
148
+ || process.env.TESTDATA_BENCHMARK_MODEL?.trim()
149
+ || '';
150
+ const modelNames = modelValue.split(',').map(value => value.trim()).filter(Boolean);
151
+ if (modelNames.length === 0) {
152
+ throw new Error('缺少环境变量 TESTDATA_BENCHMARK_MODELS 或 TESTDATA_BENCHMARK_MODEL');
153
+ }
154
+ const timeoutSeconds = Number(process.env.TESTDATA_BENCHMARK_TIMEOUT_SECONDS || 600);
155
+ if (!Number.isFinite(timeoutSeconds) || timeoutSeconds < 10 || timeoutSeconds > 1800) {
156
+ throw new Error('TESTDATA_BENCHMARK_TIMEOUT_SECONDS 必须在 10 到 1800 之间');
157
+ }
158
+ return modelNames.map((modelName, index) => ({
159
+ endpointId: `benchmark-${index + 1}`,
160
+ endpointName: 'benchmark',
161
+ apiBaseUrl,
162
+ apiKey,
163
+ modelName,
164
+ timeoutSeconds,
165
+ }));
166
+ }
167
+ async function writeJsonReport(output, payload) {
168
+ const resolved = path_1.default.resolve(output);
169
+ await (0, promises_1.mkdir)(path_1.default.dirname(resolved), { recursive: true });
170
+ await (0, promises_1.writeFile)(resolved, `${JSON.stringify(payload, null, 2)}\n`, 'utf8');
171
+ return resolved;
172
+ }
173
+ async function readJsonReport(input) {
174
+ const resolved = path_1.default.resolve(input);
175
+ let parsed;
176
+ try {
177
+ parsed = JSON.parse(await (0, promises_1.readFile)(resolved, 'utf8'));
178
+ }
179
+ catch (err) {
180
+ throw new Error(`无法读取基准报告 ${resolved}:${err instanceof Error ? err.message : String(err)}`);
181
+ }
182
+ return (0, testdataBenchmark_1.parseTestdataBenchmarkReport)(parsed);
183
+ }
184
+ function getPluginVersion() {
185
+ return String(package_json_1.default.version || process.env.npm_package_version || 'unknown');
186
+ }
187
+ function getRevision() {
188
+ const provided = process.env.TESTDATA_BENCHMARK_REVISION?.trim();
189
+ if (provided)
190
+ return provided.slice(0, 80);
191
+ try {
192
+ return (0, child_process_1.execFileSync)('git', ['rev-parse', '--short=12', 'HEAD'], {
193
+ encoding: 'utf8',
194
+ stdio: ['ignore', 'pipe', 'ignore'],
195
+ }).trim() || undefined;
196
+ }
197
+ catch {
198
+ return undefined;
199
+ }
200
+ }
201
+ async function main(argv = process.argv.slice(2)) {
202
+ let options;
203
+ try {
204
+ options = parseTestdataBenchmarkCliArgs(argv);
205
+ }
206
+ catch (err) {
207
+ console.error(err instanceof Error ? err.message : String(err));
208
+ console.error(usage());
209
+ return 2;
210
+ }
211
+ if (options.help) {
212
+ console.log(usage());
213
+ return 0;
214
+ }
215
+ if (options.list) {
216
+ for (const item of testdataBenchmarkCases_1.TESTDATA_HARD_BENCHMARK_CASES) {
217
+ console.log(`${item.id}\t${item.title}\t${item.tags.join(',')}`);
218
+ }
219
+ return 0;
220
+ }
221
+ if (options.compareReports) {
222
+ try {
223
+ const [baseline, current] = await Promise.all(options.compareReports.map(readJsonReport));
224
+ const comparison = (0, testdataBenchmark_1.compareTestdataBenchmarkReports)(baseline, current);
225
+ console.log((0, testdataBenchmark_1.formatTestdataBenchmarkComparison)(comparison));
226
+ return comparison.regressions > 0 || comparison.passRateDelta < 0 ? 1 : 0;
227
+ }
228
+ catch (err) {
229
+ console.error(err instanceof Error ? err.message : String(err));
230
+ return 2;
231
+ }
232
+ }
233
+ if (!options.confirmCost) {
234
+ console.error('未执行:真实基准会产生模型调用费用,请显式传入 --confirm-cost。');
235
+ return 2;
236
+ }
237
+ try {
238
+ const cases = (0, testdataBenchmarkCases_1.selectTestdataBenchmarkCases)(options.caseIds);
239
+ const models = buildModels();
240
+ const sandboxHost = requiredEnv('TESTDATA_BENCHMARK_SANDBOX_HOST');
241
+ const [{ MultiModelClient }, { GoJudgeSandboxRunner }] = await Promise.all([
242
+ Promise.resolve().then(() => __importStar(require('../services/openaiClient'))),
243
+ Promise.resolve().then(() => __importStar(require('../services/goJudgeSandboxService'))),
244
+ ]);
245
+ const runner = new GoJudgeSandboxRunner(sandboxHost);
246
+ if (!await runner.isAvailable()) {
247
+ throw new Error(`go-judge 不可用:${sandboxHost}`);
248
+ }
249
+ const client = new MultiModelClient(models);
250
+ const startedAt = new Date().toISOString();
251
+ const lastStage = new Map();
252
+ const results = await (0, testdataBenchmark_1.runTestdataBenchmark)(cases, client, runner, {
253
+ onCaseStart: (item, index, total) => {
254
+ console.log(`[${index + 1}/${total}] START ${item.id} ${item.title}`);
255
+ },
256
+ onProgress: (item, event) => {
257
+ const key = `${event.attempt}:${event.stage}`;
258
+ if (lastStage.get(item.id) === key)
259
+ return;
260
+ lastStage.set(item.id, key);
261
+ console.log(`[${item.id}] ${event.percent}% ${event.stage} attempt=${event.attempt}`);
262
+ },
263
+ onCaseComplete: (result) => {
264
+ console.log(`[${result.id}] ${result.passed ? 'PASS' : 'FAIL'} ${(result.durationMs / 1000).toFixed(1)}s`);
265
+ },
266
+ });
267
+ const summary = (0, testdataBenchmark_1.summarizeTestdataBenchmark)(results);
268
+ const completedAt = new Date().toISOString();
269
+ const revision = getRevision();
270
+ const report = {
271
+ schemaVersion: 1,
272
+ runId: `${completedAt.replace(/[:.]/g, '-')}-${models.map(model => model.modelName).join('+').slice(0, 80)}`,
273
+ startedAt,
274
+ completedAt,
275
+ pluginVersion: getPluginVersion(),
276
+ ...(revision ? { revision } : {}),
277
+ models: models.map(model => model.modelName),
278
+ summary,
279
+ results,
280
+ };
281
+ console.log('');
282
+ console.log((0, testdataBenchmark_1.formatTestdataBenchmarkSummary)(results));
283
+ if (options.output) {
284
+ const reportPath = await writeJsonReport(options.output, report);
285
+ console.log(`Report: ${reportPath}`);
286
+ }
287
+ if (options.aggregateOutput) {
288
+ const aggregatePath = await writeJsonReport(options.aggregateOutput, (0, testdataBenchmark_1.createTestdataBenchmarkAggregateSnapshot)(report));
289
+ console.log(`Aggregate report: ${aggregatePath}`);
290
+ }
291
+ let regressed = false;
292
+ if (options.compare) {
293
+ const baseline = await readJsonReport(options.compare);
294
+ const comparison = (0, testdataBenchmark_1.compareTestdataBenchmarkReports)(baseline, report);
295
+ console.log('');
296
+ console.log((0, testdataBenchmark_1.formatTestdataBenchmarkComparison)(comparison));
297
+ regressed = comparison.regressions > 0 || comparison.passRateDelta < 0;
298
+ }
299
+ return summary.passRate >= options.minPassRate && !regressed ? 0 : 1;
300
+ }
301
+ catch (err) {
302
+ console.error(err instanceof Error ? err.message : String(err));
303
+ return 2;
304
+ }
305
+ }
306
+ if (require.main === module) {
307
+ main().then(code => { process.exitCode = code; }).catch(err => {
308
+ console.error(err);
309
+ process.exitCode = 2;
310
+ });
311
+ }
312
+ //# sourceMappingURL=testdataBenchmarkCli.js.map
@@ -0,0 +1 @@
1
+ {"version":3,"file":"testdataBenchmarkCli.js","sourceRoot":"","sources":["../../src/benchmarks/testdataBenchmarkCli.ts"],"names":[],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AA8DA,sEAkDC;AAmED,oBAwGC;AA3RD,iDAA6C;AAC7C,0CAAyD;AACzD,gDAAwB;AACxB,sEAA6C;AAE7C,2DAS6B;AAC7B,qEAGkC;AAclC,SAAS,KAAK;IACZ,OAAO;;;;;;;;;;;;;;;;;gDAiBuC,CAAC;AACjD,CAAC;AAED,SAAS,SAAS,CAAC,IAAc,EAAE,KAAa,EAAE,IAAY;IAC5D,MAAM,OAAO,GAAG,IAAI,CAAC,KAAK,CAAC,CAAC;IAC5B,MAAM,MAAM,GAAG,GAAG,IAAI,GAAG,CAAC;IAC1B,IAAI,OAAO,CAAC,UAAU,CAAC,MAAM,CAAC;QAAE,OAAO,EAAE,KAAK,EAAE,OAAO,CAAC,KAAK,CAAC,MAAM,CAAC,MAAM,CAAC,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC;IAC5F,MAAM,IAAI,GAAG,IAAI,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC;IAC7B,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,UAAU,CAAC,IAAI,CAAC;QAAE,MAAM,IAAI,KAAK,CAAC,GAAG,IAAI,OAAO,CAAC,CAAC;IACpE,OAAO,EAAE,KAAK,EAAE,IAAI,EAAE,QAAQ,EAAE,CAAC,EAAE,CAAC;AACtC,CAAC;AAED,SAAgB,6BAA6B,CAAC,IAAc;IAC1D,MAAM,OAAO,GAAe;QAC1B,IAAI,EAAE,KAAK;QACX,IAAI,EAAE,KAAK;QACX,WAAW,EAAE,KAAK;QAClB,OAAO,EAAE,EAAE;QACX,WAAW,EAAE,CAAC;KACf,CAAC;IACF,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACrC,MAAM,GAAG,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC;QACpB,IAAI,GAAG,KAAK,QAAQ;YAAE,OAAO,CAAC,IAAI,GAAG,IAAI,CAAC;aACrC,IAAI,GAAG,KAAK,QAAQ,IAAI,GAAG,KAAK,IAAI;YAAE,OAAO,CAAC,IAAI,GAAG,IAAI,CAAC;aAC1D,IAAI,GAAG,KAAK,gBAAgB;YAAE,OAAO,CAAC,WAAW,GAAG,IAAI,CAAC;aACzD,IAAI,GAAG,KAAK,QAAQ,IAAI,GAAG,CAAC,UAAU,CAAC,SAAS,CAAC,EAAE,CAAC;YACvD,MAAM,MAAM,GAAG,SAAS,CAAC,IAAI,EAAE,CAAC,EAAE,QAAQ,CAAC,CAAC;YAC5C,OAAO,CAAC,OAAO,CAAC,IAAI,CAAC,GAAG,MAAM,CAAC,KAAK,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,CAAC;YAC5F,CAAC,IAAI,MAAM,CAAC,QAAQ,CAAC;QACvB,CAAC;aAAM,IAAI,GAAG,KAAK,UAAU,IAAI,GAAG,CAAC,UAAU,CAAC,WAAW,CAAC,EAAE,CAAC;YAC7D,MAAM,MAAM,GAAG,SAAS,CAAC,IAAI,EAAE,CAAC,EAAE,UAAU,CAAC,CAAC;YAC9C,OAAO,CAAC,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC;YAC9B,CAAC,IAAI,MAAM,CAAC,QAAQ,CAAC;QACvB,CAAC;aAAM,IAAI,GAAG,KAAK,oBAAoB,IAAI,GAAG,CAAC,UAAU,CAAC,qBAAqB,CAAC,EAAE,CAAC;YACjF,MAAM,MAAM,GAAG,SAAS,CAAC,IAAI,EAAE,CAAC,EAAE,oBAAoB,CAAC,CAAC;YACxD,OAAO,CAAC,eAAe,GAAG,MAAM,CAAC,KAAK,CAAC;YACvC,CAAC,IAAI,MAAM,CAAC,QAAQ,CAAC;QACvB,CAAC;aAAM,IAAI,GAAG,KAAK,WAAW,IAAI,GAAG,CAAC,UAAU,CAAC,YAAY,CAAC,EAAE,CAAC;YAC/D,MAAM,MAAM,GAAG,SAAS,CAAC,IAAI,EAAE,CAAC,EAAE,WAAW,CAAC,CAAC;YAC/C,OAAO,CAAC,OAAO,GAAG,MAAM,CAAC,KAAK,CAAC;YAC/B,CAAC,IAAI,MAAM,CAAC,QAAQ,CAAC;QACvB,CAAC;aAAM,IAAI,GAAG,KAAK,mBAAmB,IAAI,GAAG,CAAC,UAAU,CAAC,oBAAoB,CAAC,EAAE,CAAC;YAC/E,MAAM,MAAM,GAAG,SAAS,CAAC,IAAI,EAAE,CAAC,EAAE,mBAAmB,CAAC,CAAC;YACvD,MAAM,KAAK,GAAG,MAAM,CAAC,KAAK,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC;YACjF,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC;gBAAE,MAAM,IAAI,KAAK,CAAC,0CAA0C,CAAC,CAAC;YACpF,OAAO,CAAC,cAAc,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;YAC9C,CAAC,IAAI,MAAM,CAAC,QAAQ,CAAC;QACvB,CAAC;aAAM,IAAI,GAAG,KAAK,iBAAiB,IAAI,GAAG,CAAC,UAAU,CAAC,kBAAkB,CAAC,EAAE,CAAC;YAC3E,MAAM,MAAM,GAAG,SAAS,CAAC,IAAI,EAAE,CAAC,EAAE,iBAAiB,CAAC,CAAC;YACrD,OAAO,CAAC,WAAW,GAAG,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;YAC3C,CAAC,IAAI,MAAM,CAAC,QAAQ,CAAC;QACvB,CAAC;aAAM,CAAC;YACN,MAAM,IAAI,KAAK,CAAC,QAAQ,GAAG,EAAE,CAAC,CAAC;QACjC,CAAC;IACH,CAAC;IACD,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,OAAO,CAAC,WAAW,CAAC;WACpC,OAAO,CAAC,WAAW,GAAG,CAAC;WACvB,OAAO,CAAC,WAAW,GAAG,CAAC,EAAE,CAAC;QAC7B,MAAM,IAAI,KAAK,CAAC,iCAAiC,CAAC,CAAC;IACrD,CAAC;IACD,OAAO,CAAC,OAAO,GAAG,CAAC,GAAG,IAAI,GAAG,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC,CAAC;IAChD,OAAO,OAAO,CAAC;AACjB,CAAC;AAED,SAAS,WAAW,CAAC,IAAY;IAC/B,MAAM,KAAK,GAAG,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,CAAC;IACxC,IAAI,CAAC,KAAK;QAAE,MAAM,IAAI,KAAK,CAAC,UAAU,IAAI,EAAE,CAAC,CAAC;IAC9C,OAAO,KAAK,CAAC;AACf,CAAC;AAED,SAAS,WAAW;IAClB,MAAM,UAAU,GAAG,WAAW,CAAC,6BAA6B,CAAC,CAAC;IAC9D,MAAM,MAAM,GAAG,WAAW,CAAC,4BAA4B,CAAC,CAAC;IACzD,MAAM,UAAU,GAAG,OAAO,CAAC,GAAG,CAAC,yBAAyB,EAAE,IAAI,EAAE;WAC3D,OAAO,CAAC,GAAG,CAAC,wBAAwB,EAAE,IAAI,EAAE;WAC5C,EAAE,CAAC;IACR,MAAM,UAAU,GAAG,UAAU,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC;IACpF,IAAI,UAAU,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QAC5B,MAAM,IAAI,KAAK,CAAC,6DAA6D,CAAC,CAAC;IACjF,CAAC;IACD,MAAM,cAAc,GAAG,MAAM,CAAC,OAAO,CAAC,GAAG,CAAC,kCAAkC,IAAI,GAAG,CAAC,CAAC;IACrF,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,cAAc,CAAC,IAAI,cAAc,GAAG,EAAE,IAAI,cAAc,GAAG,IAAI,EAAE,CAAC;QACrF,MAAM,IAAI,KAAK,CAAC,qDAAqD,CAAC,CAAC;IACzE,CAAC;IACD,OAAO,UAAU,CAAC,GAAG,CAAC,CAAC,SAAS,EAAE,KAAK,EAAE,EAAE,CAAC,CAAC;QAC3C,UAAU,EAAE,aAAa,KAAK,GAAG,CAAC,EAAE;QACpC,YAAY,EAAE,WAAW;QACzB,UAAU;QACV,MAAM;QACN,SAAS;QACT,cAAc;KACf,CAAC,CAAC,CAAC;AACN,CAAC;AAED,KAAK,UAAU,eAAe,CAAC,MAAc,EAAE,OAAgB;IAC7D,MAAM,QAAQ,GAAG,cAAI,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC;IACtC,MAAM,IAAA,gBAAK,EAAC,cAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAC;IACzD,MAAM,IAAA,oBAAS,EAAC,QAAQ,EAAE,GAAG,IAAI,CAAC,SAAS,CAAC,OAAO,EAAE,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,MAAM,CAAC,CAAC;IAC3E,OAAO,QAAQ,CAAC;AAClB,CAAC;AAED,KAAK,UAAU,cAAc,CAAC,KAAa;IACzC,MAAM,QAAQ,GAAG,cAAI,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC;IACrC,IAAI,MAAe,CAAC;IACpB,IAAI,CAAC;QACH,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,MAAM,IAAA,mBAAQ,EAAC,QAAQ,EAAE,MAAM,CAAC,CAAC,CAAC;IACxD,CAAC;IAAC,OAAO,GAAG,EAAE,CAAC;QACb,MAAM,IAAI,KAAK,CAAC,YAAY,QAAQ,IAAI,GAAG,YAAY,KAAK,CAAC,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;IAC9F,CAAC;IACD,OAAO,IAAA,gDAA4B,EAAC,MAAM,CAAC,CAAC;AAC9C,CAAC;AAED,SAAS,gBAAgB;IACvB,OAAO,MAAM,CAAC,sBAAW,CAAC,OAAO,IAAI,OAAO,CAAC,GAAG,CAAC,mBAAmB,IAAI,SAAS,CAAC,CAAC;AACrF,CAAC;AAED,SAAS,WAAW;IAClB,MAAM,QAAQ,GAAG,OAAO,CAAC,GAAG,CAAC,2BAA2B,EAAE,IAAI,EAAE,CAAC;IACjE,IAAI,QAAQ;QAAE,OAAO,QAAQ,CAAC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;IAC3C,IAAI,CAAC;QACH,OAAO,IAAA,4BAAY,EAAC,KAAK,EAAE,CAAC,WAAW,EAAE,YAAY,EAAE,MAAM,CAAC,EAAE;YAC9D,QAAQ,EAAE,MAAM;YAChB,KAAK,EAAE,CAAC,QAAQ,EAAE,MAAM,EAAE,QAAQ,CAAC;SACpC,CAAC,CAAC,IAAI,EAAE,IAAI,SAAS,CAAC;IACzB,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,SAAS,CAAC;IACnB,CAAC;AACH,CAAC;AAEM,KAAK,UAAU,IAAI,CAAC,IAAI,GAAG,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC;IACrD,IAAI,OAAmB,CAAC;IACxB,IAAI,CAAC;QACH,OAAO,GAAG,6BAA6B,CAAC,IAAI,CAAC,CAAC;IAChD,CAAC;IAAC,OAAO,GAAG,EAAE,CAAC;QACb,OAAO,CAAC,KAAK,CAAC,GAAG,YAAY,KAAK,CAAC,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC;QAChE,OAAO,CAAC,KAAK,CAAC,KAAK,EAAE,CAAC,CAAC;QACvB,OAAO,CAAC,CAAC;IACX,CAAC;IACD,IAAI,OAAO,CAAC,IAAI,EAAE,CAAC;QACjB,OAAO,CAAC,GAAG,CAAC,KAAK,EAAE,CAAC,CAAC;QACrB,OAAO,CAAC,CAAC;IACX,CAAC;IACD,IAAI,OAAO,CAAC,IAAI,EAAE,CAAC;QACjB,KAAK,MAAM,IAAI,IAAI,sDAA6B,EAAE,CAAC;YACjD,OAAO,CAAC,GAAG,CAAC,GAAG,IAAI,CAAC,EAAE,KAAK,IAAI,CAAC,KAAK,KAAK,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;QACnE,CAAC;QACD,OAAO,CAAC,CAAC;IACX,CAAC;IACD,IAAI,OAAO,CAAC,cAAc,EAAE,CAAC;QAC3B,IAAI,CAAC;YACH,MAAM,CAAC,QAAQ,EAAE,OAAO,CAAC,GAAG,MAAM,OAAO,CAAC,GAAG,CAAC,OAAO,CAAC,cAAc,CAAC,GAAG,CAAC,cAAc,CAAC,CAAC,CAAC;YAC1F,MAAM,UAAU,GAAG,IAAA,mDAA+B,EAAC,QAAQ,EAAE,OAAO,CAAC,CAAC;YACtE,OAAO,CAAC,GAAG,CAAC,IAAA,qDAAiC,EAAC,UAAU,CAAC,CAAC,CAAC;YAC3D,OAAO,UAAU,CAAC,WAAW,GAAG,CAAC,IAAI,UAAU,CAAC,aAAa,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QAC5E,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC;YACb,OAAO,CAAC,KAAK,CAAC,GAAG,YAAY,KAAK,CAAC,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC;YAChE,OAAO,CAAC,CAAC;QACX,CAAC;IACH,CAAC;IACD,IAAI,CAAC,OAAO,CAAC,WAAW,EAAE,CAAC;QACzB,OAAO,CAAC,KAAK,CAAC,yCAAyC,CAAC,CAAC;QACzD,OAAO,CAAC,CAAC;IACX,CAAC;IAED,IAAI,CAAC;QACH,MAAM,KAAK,GAAG,IAAA,qDAA4B,EAAC,OAAO,CAAC,OAAO,CAAC,CAAC;QAC5D,MAAM,MAAM,GAAG,WAAW,EAAE,CAAC;QAC7B,MAAM,WAAW,GAAG,WAAW,CAAC,iCAAiC,CAAC,CAAC;QACnE,MAAM,CAAC,EAAE,gBAAgB,EAAE,EAAE,EAAE,oBAAoB,EAAE,CAAC,GAAG,MAAM,OAAO,CAAC,GAAG,CAAC;8DAClE,0BAA0B;8DAC1B,mCAAmC;SAC3C,CAAC,CAAC;QACH,MAAM,MAAM,GAAG,IAAI,oBAAoB,CAAC,WAAW,CAAC,CAAC;QACrD,IAAI,CAAC,MAAM,MAAM,CAAC,WAAW,EAAE,EAAE,CAAC;YAChC,MAAM,IAAI,KAAK,CAAC,gBAAgB,WAAW,EAAE,CAAC,CAAC;QACjD,CAAC;QACD,MAAM,MAAM,GAAG,IAAI,gBAAgB,CAAC,MAAM,CAAC,CAAC;QAC5C,MAAM,SAAS,GAAG,IAAI,IAAI,EAAE,CAAC,WAAW,EAAE,CAAC;QAC3C,MAAM,SAAS,GAAG,IAAI,GAAG,EAAkB,CAAC;QAC5C,MAAM,OAAO,GAAG,MAAM,IAAA,wCAAoB,EAAC,KAAK,EAAE,MAAM,EAAE,MAAM,EAAE;YAChE,WAAW,EAAE,CAAC,IAAI,EAAE,KAAK,EAAE,KAAK,EAAE,EAAE;gBAClC,OAAO,CAAC,GAAG,CAAC,IAAI,KAAK,GAAG,CAAC,IAAI,KAAK,WAAW,IAAI,CAAC,EAAE,IAAI,IAAI,CAAC,KAAK,EAAE,CAAC,CAAC;YACxE,CAAC;YACD,UAAU,EAAE,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE;gBAC1B,MAAM,GAAG,GAAG,GAAG,KAAK,CAAC,OAAO,IAAI,KAAK,CAAC,KAAK,EAAE,CAAC;gBAC9C,IAAI,SAAS,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,KAAK,GAAG;oBAAE,OAAO;gBAC3C,SAAS,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,EAAE,GAAG,CAAC,CAAC;gBAC5B,OAAO,CAAC,GAAG,CAAC,IAAI,IAAI,CAAC,EAAE,KAAK,KAAK,CAAC,OAAO,KAAK,KAAK,CAAC,KAAK,YAAY,KAAK,CAAC,OAAO,EAAE,CAAC,CAAC;YACxF,CAAC;YACD,cAAc,EAAE,CAAC,MAAM,EAAE,EAAE;gBACzB,OAAO,CAAC,GAAG,CAAC,IAAI,MAAM,CAAC,EAAE,KAAK,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,IAAI,CAAC,MAAM,CAAC,UAAU,GAAG,IAAI,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC;YAC7G,CAAC;SACF,CAAC,CAAC;QACH,MAAM,OAAO,GAAG,IAAA,8CAA0B,EAAC,OAAO,CAAC,CAAC;QACpD,MAAM,WAAW,GAAG,IAAI,IAAI,EAAE,CAAC,WAAW,EAAE,CAAC;QAC7C,MAAM,QAAQ,GAAG,WAAW,EAAE,CAAC;QAC/B,MAAM,MAAM,GAA4B;YACtC,aAAa,EAAE,CAAC;YAChB,KAAK,EAAE,GAAG,WAAW,CAAC,OAAO,CAAC,OAAO,EAAE,GAAG,CAAC,IAAI,MAAM,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC,KAAK,CAAC,SAAS,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,EAAE;YAC5G,SAAS;YACT,WAAW;YACX,aAAa,EAAE,gBAAgB,EAAE;YACjC,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,QAAQ,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;YACjC,MAAM,EAAE,MAAM,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC,KAAK,CAAC,SAAS,CAAC;YAC5C,OAAO;YACP,OAAO;SACR,CAAC;QACF,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;QAChB,OAAO,CAAC,GAAG,CAAC,IAAA,kDAA8B,EAAC,OAAO,CAAC,CAAC,CAAC;QACrD,IAAI,OAAO,CAAC,MAAM,EAAE,CAAC;YACnB,MAAM,UAAU,GAAG,MAAM,eAAe,CAAC,OAAO,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;YACjE,OAAO,CAAC,GAAG,CAAC,WAAW,UAAU,EAAE,CAAC,CAAC;QACvC,CAAC;QACD,IAAI,OAAO,CAAC,eAAe,EAAE,CAAC;YAC5B,MAAM,aAAa,GAAG,MAAM,eAAe,CACzC,OAAO,CAAC,eAAe,EACvB,IAAA,4DAAwC,EAAC,MAAM,CAAC,CACjD,CAAC;YACF,OAAO,CAAC,GAAG,CAAC,qBAAqB,aAAa,EAAE,CAAC,CAAC;QACpD,CAAC;QACD,IAAI,SAAS,GAAG,KAAK,CAAC;QACtB,IAAI,OAAO,CAAC,OAAO,EAAE,CAAC;YACpB,MAAM,QAAQ,GAAG,MAAM,cAAc,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC;YACvD,MAAM,UAAU,GAAG,IAAA,mDAA+B,EAAC,QAAQ,EAAE,MAAM,CAAC,CAAC;YACrE,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;YAChB,OAAO,CAAC,GAAG,CAAC,IAAA,qDAAiC,EAAC,UAAU,CAAC,CAAC,CAAC;YAC3D,SAAS,GAAG,UAAU,CAAC,WAAW,GAAG,CAAC,IAAI,UAAU,CAAC,aAAa,GAAG,CAAC,CAAC;QACzE,CAAC;QACD,OAAO,OAAO,CAAC,QAAQ,IAAI,OAAO,CAAC,WAAW,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IACvE,CAAC;IAAC,OAAO,GAAG,EAAE,CAAC;QACb,OAAO,CAAC,KAAK,CAAC,GAAG,YAAY,KAAK,CAAC,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC;QAChE,OAAO,CAAC,CAAC;IACX,CAAC;AACH,CAAC;AAED,IAAI,OAAO,CAAC,IAAI,KAAK,MAAM,EAAE,CAAC;IAC5B,IAAI,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,GAAG,OAAO,CAAC,QAAQ,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE;QAC5D,OAAO,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;QACnB,OAAO,CAAC,QAAQ,GAAG,CAAC,CAAC;IACvB,CAAC,CAAC,CAAC;AACL,CAAC"}
@@ -1,47 +1,63 @@
1
1
  "use strict";
2
2
  /**
3
- * 系统内置的越狱关键词/正则
3
+ * 系统内置的安全策略规则。
4
+ *
5
+ * “直接索要答案”与真正的 Prompt Injection 分开分类,避免教学策略违规
6
+ * 被误标为安全攻击。规则只是第一层确定性检测,不能替代输出门禁。
4
7
  */
5
8
  Object.defineProperty(exports, "__esModule", { value: true });
6
- exports.getBuiltinJailbreakPatterns = exports.builtinJailbreakPatternSources = void 0;
7
- const BUILTIN_PATTERNS = [
8
- /忽略(之前|上文|所有).*提示/gi,
9
- /ignore (all|previous|earlier) (instructions|messages|prompts)/gi,
10
- /(从现在开始|现在起?).*(你是|扮演).*(猫娘|女仆|主人|角色|人格)/gi,
11
- /重置(设定|設定|系统|system)/gi,
12
- /system prompt/gi,
13
- /无条件服从/gi,
14
- /覆盖(系统|所有)提示/gi,
15
- /现在你是.*系统/gi,
16
- /(前面|上述|上面).*(规则|说明|内容).*(视为|看作).*(参考|示例|背景)/gi,
17
- /(这些|那些).*(规则|要求).*(只是|仅仅是).*(参考|建议)/gi,
18
- /(后续|接下来|在.*回答中).*(遇到|如果).*(冲突|不一致).*(优先|首先).*(执行|遵守).*(当前|本段|下面|以下).*(规则|内容)/gi,
19
- /如果.*(冲突|矛盾).*(请|就).*(以.*(为准|为先|为主))/gi,
20
- /(满足|服从).*(用户|我).*(最新|当前).*(请求|需求).*(视为|看作).*(最高优先级|第一优先)/gi,
21
- /(treat|treated)\s+as\s+(reference|background)(\s+only)?/gi,
22
- /(this|the following).*(takes?|take)\s+precedence/gi,
23
- /in\s+case\s+of\s+conflict.*(follow|use)\s+(this|the following)/gi,
24
- /user('?s)?\s+latest\s+(request|instruction).*(highest\s+priority)/gi,
9
+ exports.getBuiltinJailbreakPatterns = exports.getBuiltinJailbreakRules = exports.builtinJailbreakPatternSources = void 0;
10
+ const INJECTION_RULES = [
11
+ { pattern: /忽略(之前|上文|所有).*提示/gi, category: 'prompt_injection', riskScore: 85 },
12
+ { pattern: /无视(之前|上文|前面|所有).*(要求|规则|提示|设定)/gi, category: 'prompt_injection', riskScore: 85 },
13
+ { pattern: /ignore\s+(all\s+)?(previous|earlier)\s+(instructions|messages|prompts)/gi, category: 'prompt_injection', riskScore: 85 },
14
+ // 紧凑视图用于识别 i g n o r e / 忽 略 一类字符插空绕过。
15
+ { pattern: /ignore(?:all)?(?:previous|earlier)(?:instructions|messages|prompts)/gi, category: 'prompt_injection', riskScore: 90 },
16
+ { pattern: /(从现在开始|现在起?).*(你是|扮演).*(猫娘|女仆|主人|角色|人格)/gi, category: 'prompt_injection', riskScore: 80 },
17
+ { pattern: /重置(设定|設定|系统|system)/gi, category: 'prompt_injection', riskScore: 85 },
18
+ { pattern: /无条件服从/gi, category: 'prompt_injection', riskScore: 80 },
19
+ { pattern: /覆盖(系统|所有)提示/gi, category: 'prompt_injection', riskScore: 90 },
20
+ { pattern: /现在你是.*系统/gi, category: 'prompt_injection', riskScore: 85 },
21
+ { pattern: /(前面|上述|上面).*(规则|说明|内容).*(视为|看作).*(参考|示例|背景)/gi, category: 'prompt_injection', riskScore: 80 },
22
+ { pattern: /(这些|那些).*(规则|要求).*(只是|仅仅是).*(参考|建议)/gi, category: 'prompt_injection', riskScore: 80 },
23
+ { pattern: /(后续|接下来|在.*回答中).*(遇到|如果).*(冲突|不一致).*(优先|首先).*(执行|遵守).*(当前|本段|下面|以下).*(规则|内容)/gi, category: 'prompt_injection', riskScore: 85 },
24
+ { pattern: /如果.*(冲突|矛盾).*(请|就).*(以.*(为准|为先|为主))/gi, category: 'prompt_injection', riskScore: 80 },
25
+ { pattern: /(满足|服从).*(用户|我).*(最新|当前).*(请求|需求).*(视为|看作).*(最高优先级|第一优先)/gi, category: 'prompt_injection', riskScore: 85 },
26
+ { pattern: /(treat|treated)\s+as\s+(reference|background)(\s+only)?/gi, category: 'prompt_injection', riskScore: 75 },
27
+ { pattern: /(this|the following).*(takes?|take)\s+precedence/gi, category: 'prompt_injection', riskScore: 80 },
28
+ { pattern: /in\s+case\s+of\s+conflict.*(follow|use)\s+(this|the following)/gi, category: 'prompt_injection', riskScore: 85 },
29
+ { pattern: /user('?s)?\s+latest\s+(request|instruction).*(highest\s+priority)/gi, category: 'prompt_injection', riskScore: 85 },
30
+ { pattern: /\bDAN\b/g, category: 'prompt_injection', riskScore: 80 },
31
+ { pattern: /\bdo anything now\b/gi, category: 'prompt_injection', riskScore: 85 },
32
+ { pattern: /developer mode/gi, category: 'prompt_injection', riskScore: 80 },
33
+ { pattern: /\bjailbreak\b/gi, category: 'prompt_injection', riskScore: 75 },
34
+ { pattern: /\[(AI导师|系统|assistant)\]\s*[::]/gi, category: 'prompt_injection', riskScore: 80 },
35
+ ];
36
+ const ANSWER_SEEKING_RULES = [
25
37
  // 直接索要答案
26
- /直接给[我出]?.*?(答案|代码|solution)/gi,
27
- /给我完整.*?代码/gi,
28
- /不要.*?教学模式/gi,
29
- /把(完整|全部).*?(代码|答案|solution).*?(给|发|写|输出)/gi,
30
- // DAN / 越狱
31
- /\bDAN\b/g,
32
- /\bdo anything now\b/gi,
33
- /developer mode/gi,
34
- /\bjailbreak\b/gi,
35
- // 角色标签伪造
36
- /\[(AI导师|系统|assistant)\]\s*[::]/gi,
37
- // 编码绕过
38
- /(base64|decode|eval)\s*[::((]/gi,
39
- /解码.*?以下/gi,
40
- // System prompt 泄露
41
- /输出.*?系统.*?提示词/gi,
42
- /(show|print|output|reveal).*?system.*?prompt/gi,
38
+ { pattern: /直接给[我出]?.*?(答案|代码|solution)/gi, category: 'answer_seeking', riskScore: 40 },
39
+ { pattern: /给我完整.*?代码/gi, category: 'answer_seeking', riskScore: 40 },
40
+ { pattern: /不要.*?教学模式/gi, category: 'answer_seeking', riskScore: 45 },
41
+ { pattern: /把(完整|全部).*?(代码|答案|solution).*?(给|发|写|输出)/gi, category: 'answer_seeking', riskScore: 45 },
42
+ { pattern: /(写出|给出|生成).*(可通过|通过).*(全部|所有).*(测试点|样例).*(代码|程序|实现)/gi, category: 'answer_seeking', riskScore: 45 },
43
+ { pattern: /(accepted|complete|full)\s+(solution|code|implementation)/gi, category: 'answer_seeking', riskScore: 40 },
44
+ ];
45
+ const EXFILTRATION_RULES = [
46
+ { pattern: /(输出|展示|泄露|告诉我|打印).*?(系统|system).*?(提示词|prompt|指令)/gi, category: 'prompt_exfiltration', riskScore: 90 },
47
+ { pattern: /(show|print|output|reveal|repeat).*?(your|the)?.*?system.*?(prompt|instructions)/gi, category: 'prompt_exfiltration', riskScore: 90 },
48
+ { pattern: /(?:show|print|output|reveal)(?:me)?(?:your|the)?system(?:prompt|instructions)/gi, category: 'prompt_exfiltration', riskScore: 95 },
49
+ ];
50
+ const BUILTIN_RULES = [
51
+ ...INJECTION_RULES,
52
+ ...ANSWER_SEEKING_RULES,
53
+ ...EXFILTRATION_RULES,
43
54
  ];
44
- exports.builtinJailbreakPatternSources = BUILTIN_PATTERNS.map((pattern) => pattern.source);
45
- const getBuiltinJailbreakPatterns = () => BUILTIN_PATTERNS.map((pattern) => new RegExp(pattern.source, pattern.flags));
55
+ exports.builtinJailbreakPatternSources = BUILTIN_RULES.map((rule) => rule.pattern.source);
56
+ const getBuiltinJailbreakRules = () => BUILTIN_RULES.map((rule) => ({
57
+ ...rule,
58
+ pattern: new RegExp(rule.pattern.source, rule.pattern.flags),
59
+ }));
60
+ exports.getBuiltinJailbreakRules = getBuiltinJailbreakRules;
61
+ const getBuiltinJailbreakPatterns = () => (0, exports.getBuiltinJailbreakRules)().map((rule) => rule.pattern);
46
62
  exports.getBuiltinJailbreakPatterns = getBuiltinJailbreakPatterns;
47
63
  //# sourceMappingURL=jailbreakRules.js.map
@@ -1 +1 @@
1
- {"version":3,"file":"jailbreakRules.js","sourceRoot":"","sources":["../../src/constants/jailbreakRules.ts"],"names":[],"mappings":";AAAA;;GAEG;;;AAEH,MAAM,gBAAgB,GAAa;IACjC,oBAAoB;IACpB,iEAAiE;IACjE,2CAA2C;IAC3C,uBAAuB;IACvB,iBAAiB;IACjB,SAAS;IACT,eAAe;IACf,YAAY;IACZ,+CAA+C;IAC/C,uCAAuC;IACvC,gFAAgF;IAChF,uCAAuC;IACvC,4DAA4D;IAC5D,2DAA2D;IAC3D,oDAAoD;IACpD,kEAAkE;IAClE,qEAAqE;IACrE,SAAS;IACT,+BAA+B;IAC/B,aAAa;IACb,aAAa;IACb,4CAA4C;IAC5C,WAAW;IACX,UAAU;IACV,uBAAuB;IACvB,kBAAkB;IAClB,iBAAiB;IACjB,SAAS;IACT,kCAAkC;IAClC,OAAO;IACP,iCAAiC;IACjC,WAAW;IACX,mBAAmB;IACnB,iBAAiB;IACjB,gDAAgD;CACjD,CAAC;AAEW,QAAA,8BAA8B,GAAG,gBAAgB,CAAC,GAAG,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC;AAEzF,MAAM,2BAA2B,GAAG,GAAa,EAAE,CACxD,gBAAgB,CAAC,GAAG,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,MAAM,EAAE,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC;AADlE,QAAA,2BAA2B,+BACuC"}
1
+ {"version":3,"file":"jailbreakRules.js","sourceRoot":"","sources":["../../src/constants/jailbreakRules.ts"],"names":[],"mappings":";AAAA;;;;;GAKG;;;AAIH,MAAM,eAAe,GAA2B;IAC9C,EAAE,OAAO,EAAE,oBAAoB,EAAE,QAAQ,EAAE,kBAAkB,EAAE,SAAS,EAAE,EAAE,EAAE;IAC9E,EAAE,OAAO,EAAE,kCAAkC,EAAE,QAAQ,EAAE,kBAAkB,EAAE,SAAS,EAAE,EAAE,EAAE;IAC5F,EAAE,OAAO,EAAE,0EAA0E,EAAE,QAAQ,EAAE,kBAAkB,EAAE,SAAS,EAAE,EAAE,EAAE;IACpI,uCAAuC;IACvC,EAAE,OAAO,EAAE,uEAAuE,EAAE,QAAQ,EAAE,kBAAkB,EAAE,SAAS,EAAE,EAAE,EAAE;IACjI,EAAE,OAAO,EAAE,2CAA2C,EAAE,QAAQ,EAAE,kBAAkB,EAAE,SAAS,EAAE,EAAE,EAAE;IACrG,EAAE,OAAO,EAAE,uBAAuB,EAAE,QAAQ,EAAE,kBAAkB,EAAE,SAAS,EAAE,EAAE,EAAE;IACjF,EAAE,OAAO,EAAE,SAAS,EAAE,QAAQ,EAAE,kBAAkB,EAAE,SAAS,EAAE,EAAE,EAAE;IACnE,EAAE,OAAO,EAAE,eAAe,EAAE,QAAQ,EAAE,kBAAkB,EAAE,SAAS,EAAE,EAAE,EAAE;IACzE,EAAE,OAAO,EAAE,YAAY,EAAE,QAAQ,EAAE,kBAAkB,EAAE,SAAS,EAAE,EAAE,EAAE;IACtE,EAAE,OAAO,EAAE,+CAA+C,EAAE,QAAQ,EAAE,kBAAkB,EAAE,SAAS,EAAE,EAAE,EAAE;IACzG,EAAE,OAAO,EAAE,uCAAuC,EAAE,QAAQ,EAAE,kBAAkB,EAAE,SAAS,EAAE,EAAE,EAAE;IACjG,EAAE,OAAO,EAAE,gFAAgF,EAAE,QAAQ,EAAE,kBAAkB,EAAE,SAAS,EAAE,EAAE,EAAE;IAC1I,EAAE,OAAO,EAAE,uCAAuC,EAAE,QAAQ,EAAE,kBAAkB,EAAE,SAAS,EAAE,EAAE,EAAE;IACjG,EAAE,OAAO,EAAE,4DAA4D,EAAE,QAAQ,EAAE,kBAAkB,EAAE,SAAS,EAAE,EAAE,EAAE;IACtH,EAAE,OAAO,EAAE,2DAA2D,EAAE,QAAQ,EAAE,kBAAkB,EAAE,SAAS,EAAE,EAAE,EAAE;IACrH,EAAE,OAAO,EAAE,oDAAoD,EAAE,QAAQ,EAAE,kBAAkB,EAAE,SAAS,EAAE,EAAE,EAAE;IAC9G,EAAE,OAAO,EAAE,kEAAkE,EAAE,QAAQ,EAAE,kBAAkB,EAAE,SAAS,EAAE,EAAE,EAAE;IAC5H,EAAE,OAAO,EAAE,qEAAqE,EAAE,QAAQ,EAAE,kBAAkB,EAAE,SAAS,EAAE,EAAE,EAAE;IAC/H,EAAE,OAAO,EAAE,UAAU,EAAE,QAAQ,EAAE,kBAAkB,EAAE,SAAS,EAAE,EAAE,EAAE;IACpE,EAAE,OAAO,EAAE,uBAAuB,EAAE,QAAQ,EAAE,kBAAkB,EAAE,SAAS,EAAE,EAAE,EAAE;IACjF,EAAE,OAAO,EAAE,kBAAkB,EAAE,QAAQ,EAAE,kBAAkB,EAAE,SAAS,EAAE,EAAE,EAAE;IAC5E,EAAE,OAAO,EAAE,iBAAiB,EAAE,QAAQ,EAAE,kBAAkB,EAAE,SAAS,EAAE,EAAE,EAAE;IAC3E,EAAE,OAAO,EAAE,kCAAkC,EAAE,QAAQ,EAAE,kBAAkB,EAAE,SAAS,EAAE,EAAE,EAAE;CAC7F,CAAC;AAEF,MAAM,oBAAoB,GAA2B;IACnD,SAAS;IACT,EAAE,OAAO,EAAE,+BAA+B,EAAE,QAAQ,EAAE,gBAAgB,EAAE,SAAS,EAAE,EAAE,EAAE;IACvF,EAAE,OAAO,EAAE,aAAa,EAAE,QAAQ,EAAE,gBAAgB,EAAE,SAAS,EAAE,EAAE,EAAE;IACrE,EAAE,OAAO,EAAE,aAAa,EAAE,QAAQ,EAAE,gBAAgB,EAAE,SAAS,EAAE,EAAE,EAAE;IACrE,EAAE,OAAO,EAAE,4CAA4C,EAAE,QAAQ,EAAE,gBAAgB,EAAE,SAAS,EAAE,EAAE,EAAE;IACpG,EAAE,OAAO,EAAE,uDAAuD,EAAE,QAAQ,EAAE,gBAAgB,EAAE,SAAS,EAAE,EAAE,EAAE;IAC/G,EAAE,OAAO,EAAE,6DAA6D,EAAE,QAAQ,EAAE,gBAAgB,EAAE,SAAS,EAAE,EAAE,EAAE;CACtH,CAAC;AAEF,MAAM,kBAAkB,GAA2B;IACjD,EAAE,OAAO,EAAE,qDAAqD,EAAE,QAAQ,EAAE,qBAAqB,EAAE,SAAS,EAAE,EAAE,EAAE;IAClH,EAAE,OAAO,EAAE,oFAAoF,EAAE,QAAQ,EAAE,qBAAqB,EAAE,SAAS,EAAE,EAAE,EAAE;IACjJ,EAAE,OAAO,EAAE,iFAAiF,EAAE,QAAQ,EAAE,qBAAqB,EAAE,SAAS,EAAE,EAAE,EAAE;CAC/I,CAAC;AAEF,MAAM,aAAa,GAA2B;IAC5C,GAAG,eAAe;IAClB,GAAG,oBAAoB;IACvB,GAAG,kBAAkB;CACtB,CAAC;AAEW,QAAA,8BAA8B,GAAG,aAAa,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC;AAExF,MAAM,wBAAwB,GAAG,GAA2B,EAAE,CACnE,aAAa,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC;IAC3B,GAAG,IAAI;IACP,OAAO,EAAE,IAAI,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,IAAI,CAAC,OAAO,CAAC,KAAK,CAAC;CAC7D,CAAC,CAAC,CAAC;AAJO,QAAA,wBAAwB,4BAI/B;AAEC,MAAM,2BAA2B,GAAG,GAAa,EAAE,CACxD,IAAA,gCAAwB,GAAE,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;AAD5C,QAAA,2BAA2B,+BACiB"}