@saulwade/swl-ses 2.6.1 → 2.8.0

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 (268) hide show
  1. package/CLAUDE.md +14 -2
  2. package/README.md +65 -18
  3. package/agentes/_intent-spec.md +73 -73
  4. package/agentes/_propose-step.md +90 -90
  5. package/bin/swl-ses.js +10 -0
  6. package/comandos/swl/brainstorm.md +1 -0
  7. package/comandos/swl/briefing.md +119 -119
  8. package/comandos/swl/contribuir.md +233 -233
  9. package/comandos/swl/deuda-codigo.md +97 -97
  10. package/comandos/swl/mcp-status.md +1 -0
  11. package/gateway/lib/event-channel.js +191 -191
  12. package/habilidades/agent-deep-links/SKILL.md +148 -148
  13. package/habilidades/backend-async-postgres-testing/SKILL.md +216 -216
  14. package/habilidades/backend-error-design/SKILL.md +221 -221
  15. package/habilidades/backend-production-resilience/SKILL.md +288 -288
  16. package/habilidades/calidad-anti-patrones-universales/SKILL.md +105 -1
  17. package/habilidades/calidad-contract-testing/SKILL.md +165 -165
  18. package/habilidades/calidad-mutation-testing/SKILL.md +25 -1
  19. package/habilidades/checklist-seguridad/recursos/stride-cobertura.md +60 -60
  20. package/habilidades/ci-cd-pipelines/SKILL.md +5 -1
  21. package/habilidades/css-moderno/SKILL.md +7 -1
  22. package/habilidades/diagrama-arquitectura/assets/template.html +276 -276
  23. package/habilidades/doubt-driven-review/recursos/EXAMPLES.md +130 -130
  24. package/habilidades/estructura-proyecto-claude/recursos/mcp-json-template.json +57 -57
  25. package/habilidades/extractor-de-aprendizajes/SKILL.md +5 -1
  26. package/habilidades/feynman-auditor-swl/recursos/preguntas-language-agnostic.md +108 -108
  27. package/habilidades/harness-claude-code/SKILL.md +3 -2
  28. package/habilidades/meta-skills-estandar/recursos/convencion-examples.md +93 -93
  29. package/habilidades/patrones-python/recursos/patrones-avanzados.md +469 -469
  30. package/habilidades/perfil-usuario/SKILL.md +200 -200
  31. package/habilidades/prevencion-sobreingenieria/recursos/EXAMPLES.md +580 -580
  32. package/habilidades/prevencion-sobreingenieria/recursos/soluciones-nativas.md +166 -166
  33. package/habilidades/prevencion-sobreingenieria/recursos/variables-residuales-post-refactor.md +85 -85
  34. package/habilidades/proceso-ddia-streaming/SKILL.md +231 -231
  35. package/habilidades/proceso-discovery-machote/SKILL.md +157 -157
  36. package/habilidades/proceso-dynamic-workflows/SKILL.md +60 -0
  37. package/habilidades/proceso-dynamic-workflows/recursos/template-adversarial-verify.js +65 -65
  38. package/habilidades/proceso-dynamic-workflows/recursos/template-triage.js +65 -65
  39. package/habilidades/proceso-intent-engineering/SKILL.md +269 -269
  40. package/habilidades/proceso-modular-split/SKILL.md +256 -256
  41. package/habilidades/state-inconsistency-auditor-swl/recursos/coupled-state-patterns.md +147 -147
  42. package/habilidades/swl-claudemd/recursos/contrato-aprender.md +83 -83
  43. package/habilidades/swl-claudemd/recursos/duplicacion-reglas-globales.md +85 -85
  44. package/habilidades/swl-claudemd/recursos/plantillas-init.md +94 -94
  45. package/habilidades/tdd-workflow/recursos/gherkin-bdd.md +111 -111
  46. package/hooks/calidad-pre-commit.js +159 -10
  47. package/hooks/ciclo-evolucion-subagente.js +26 -26
  48. package/hooks/ciclo-evolucion.js +26 -26
  49. package/hooks/contexto-subagente.js +68 -68
  50. package/hooks/lib/auto-consolidator.js +335 -335
  51. package/hooks/lib/ciclo-evolucion.js +47 -47
  52. package/hooks/lib/deep-links.js +185 -185
  53. package/hooks/lib/error-classifier.js +308 -308
  54. package/hooks/lib/notificacion-formato.js +45 -11
  55. package/hooks/lib/provenance-tracker.js +191 -191
  56. package/hooks/lib/raiz-proyecto.js +35 -4
  57. package/hooks/lib/resource-quota.js +122 -122
  58. package/hooks/lib/retry-jitter.js +165 -165
  59. package/hooks/lib/security-net.js +201 -201
  60. package/hooks/lib/skill-auditor.js +588 -588
  61. package/hooks/lib/sync-status.js +228 -228
  62. package/hooks/lib/taint-tracker.js +107 -107
  63. package/hooks/lib/text-similarity.js +241 -241
  64. package/hooks/lib/toon-compressor.js +245 -245
  65. package/hooks/notificacion-telegram.js +5 -11
  66. package/hooks/session-briefing.js +12 -4
  67. package/instintos/autonomia.yaml +27 -27
  68. package/instintos/prompt-appendices.yaml +57 -57
  69. package/llms.txt +1 -1
  70. package/manifiestos/agent-output-schemas.json +57 -57
  71. package/manifiestos/canonical-hashes.json +662 -0
  72. package/manifiestos/harness-ir.json +47536 -0
  73. package/manifiestos/hooks-config.json +469 -469
  74. package/manifiestos/invariantes-criticos.json +30 -30
  75. package/manifiestos/policy-bundle.json +2065 -0
  76. package/manifiestos/policy-corpus-w2.json +3926 -0
  77. package/manifiestos/runtime-adapters-core3.json +208 -0
  78. package/manifiestos/runtime-conformance.json +139 -0
  79. package/manifiestos/skills-lock.json +43 -43
  80. package/package.json +2 -2
  81. package/plantillas/auditor-veto-template.md +105 -105
  82. package/plantillas/github-workflows/release-please.yml +44 -44
  83. package/plantillas/github-workflows/swl-ci.yml +107 -107
  84. package/plantillas/github-workflows/swl-security.yml +51 -51
  85. package/plugin.json +2 -2
  86. package/reglas/accesibilidad.md +10 -10
  87. package/reglas/auditorias-documentales-estructurales.md +7 -7
  88. package/reglas/cloud-infra.md +8 -8
  89. package/reglas/consultar-vault-primero.md +195 -195
  90. package/reglas/git-workflow.md +1 -0
  91. package/reglas/hooks.md +6 -6
  92. package/reglas/intent-engineering.md +218 -218
  93. package/reglas/markitdown.md +8 -8
  94. package/reglas/monitor-ci.md +12 -0
  95. package/reglas/patrones.md +6 -6
  96. package/reglas/testing.md +7 -7
  97. package/reglas/tests-cleanup.md +224 -224
  98. package/schemas/agent-message.schema.json +73 -73
  99. package/schemas/agent-output-implementacion.schema.json +114 -114
  100. package/schemas/agent-output-planificacion.schema.json +150 -150
  101. package/schemas/agent-output-review.schema.json +98 -98
  102. package/schemas/diary-entry.schema.json +112 -112
  103. package/schemas/gate-state.schema.json +76 -0
  104. package/schemas/harness-ir.schema.json +369 -0
  105. package/schemas/hook-profiles.schema.json +54 -54
  106. package/schemas/hooks-config.schema.json +89 -89
  107. package/schemas/legacy-gates.schema.json +45 -0
  108. package/schemas/modulos.schema.json +38 -38
  109. package/schemas/perfiles.schema.json +36 -36
  110. package/schemas/plugin.schema.json +77 -77
  111. package/schemas/policy-bundle.schema.json +140 -0
  112. package/schemas/policy-enforcement.schema.json +117 -0
  113. package/schemas/policy-operation.schema.json +261 -0
  114. package/schemas/runtime-adapter.schema.json +176 -0
  115. package/schemas/runtime-build-attestation.schema.json +100 -0
  116. package/schemas/runtime-conformance.schema.json +239 -0
  117. package/schemas/runtime-diagnostic.schema.json +395 -0
  118. package/schemas/skill-evals.schema.json +119 -119
  119. package/schemas/skill-frontmatter.schema.json +245 -245
  120. package/schemas/w4-certification-request.schema.json +72 -0
  121. package/schemas/w4-certification-verdict.schema.json +224 -0
  122. package/schemas/w4-corpus.schema.json +172 -0
  123. package/schemas/w4-mutation-report.schema.json +116 -0
  124. package/schemas/w4-replay-result.schema.json +164 -0
  125. package/schemas/w4-scoring-report.schema.json +89 -0
  126. package/scripts/audit-tools/audit-history.js +330 -330
  127. package/scripts/audit-tools/bundle-tracker.js +290 -290
  128. package/scripts/audit-tools/canary-monitor.js +352 -352
  129. package/scripts/audit-tools/code-profiler.js +605 -605
  130. package/scripts/audit-tools/dep-doctor.js +320 -320
  131. package/scripts/audit-tools/env-validator.js +206 -206
  132. package/scripts/audit-tools/lib/fs-walk.js +48 -48
  133. package/scripts/audit-tools/lib/output.js +23 -23
  134. package/scripts/audit-tools/migration-checker.js +392 -392
  135. package/scripts/audit-tools/pentest-scanner.js +1436 -1436
  136. package/scripts/bootstrap-instintos.js +3 -0
  137. package/scripts/cli/aprobar-plan.js +73 -73
  138. package/scripts/cli/briefing.js +23 -23
  139. package/scripts/cli/ciclo-evolucion.js +26 -26
  140. package/scripts/cli/derivar-feature-list.js +25 -25
  141. package/scripts/cli/detectar-host.js +27 -27
  142. package/scripts/cli/diary-entry.js +69 -69
  143. package/scripts/cli/execution-state.js +18 -18
  144. package/scripts/cli/gateway-notify.js +41 -41
  145. package/scripts/cli/liberar-fase.js +42 -42
  146. package/scripts/cli/mark-evolved.js +56 -56
  147. package/scripts/cli/metricas-dora.js +26 -26
  148. package/scripts/cli/near-duplicate.js +55 -55
  149. package/scripts/cli/notificaciones.js +123 -123
  150. package/scripts/cli/propose-step.js +29 -29
  151. package/scripts/cli/schedule-parse.js +19 -19
  152. package/scripts/cli/sugerir-modelo.js +20 -20
  153. package/scripts/cli/verificar-plan.js +36 -36
  154. package/scripts/cli/verificar-trazabilidad.js +35 -35
  155. package/scripts/comandos/install-asistido.js +8 -7
  156. package/scripts/configurar-branch-protection.js +418 -418
  157. package/scripts/detectar-aprendizajes-duplicados.js +151 -151
  158. package/scripts/doctor.js +61 -36
  159. package/scripts/generar-checklists-consolidados.js +273 -273
  160. package/scripts/generar-claims-runtime.js +1342 -0
  161. package/scripts/generar-harness-ir.js +257 -0
  162. package/scripts/generar-inventario.js +52 -54
  163. package/scripts/generar-policy-bundle.js +202 -0
  164. package/scripts/instalador.js +26 -7
  165. package/scripts/lib/approval-receipts.js +190 -0
  166. package/scripts/lib/artefactos-python.js +43 -43
  167. package/scripts/lib/benchmark-metrics.js +160 -160
  168. package/scripts/lib/budget-enforcer.js +252 -252
  169. package/scripts/lib/certificacion-loop-state.js +421 -0
  170. package/scripts/lib/ci-reader.js +193 -193
  171. package/scripts/lib/ciclo-autonomo/yaml-instintos.js +56 -0
  172. package/scripts/lib/clasificar-directorio.js +92 -0
  173. package/scripts/lib/contadores-inventario.js +217 -217
  174. package/scripts/lib/detectar-host-swl.js +175 -175
  175. package/scripts/lib/detectar-runtime.js +29 -20
  176. package/scripts/lib/detectar-stack-detallado.js +307 -307
  177. package/scripts/lib/detector-autoduplicacion-intra-archivo.js +234 -234
  178. package/scripts/lib/detector-reglas-duplicadas.js +220 -220
  179. package/scripts/lib/eval-metrics-store.js +218 -218
  180. package/scripts/lib/eval-quality.js +171 -171
  181. package/scripts/lib/eval-schemas.js +144 -144
  182. package/scripts/lib/eval-self-correct.js +106 -106
  183. package/scripts/lib/eval-validator.js +185 -185
  184. package/scripts/lib/evidence-verifier.js +192 -0
  185. package/scripts/lib/evidencia-release.js +322 -322
  186. package/scripts/lib/frontmatter-canonico.js +509 -0
  187. package/scripts/lib/gate-engine.js +871 -0
  188. package/scripts/lib/gate-hooks-requires.js +249 -249
  189. package/scripts/lib/gate-licencias.js +212 -212
  190. package/scripts/lib/git-config-preflight.js +48 -0
  191. package/scripts/lib/git-metricas.js +257 -257
  192. package/scripts/lib/harness-ir.js +778 -0
  193. package/scripts/lib/harness-source-snapshot.js +309 -0
  194. package/scripts/lib/integrity-ledger.js +1147 -0
  195. package/scripts/lib/jaccard-similarity.js +98 -98
  196. package/scripts/lib/legacy-gate-migration.js +324 -0
  197. package/scripts/lib/limpiar-basura-global.js +45 -2
  198. package/scripts/lib/longmemeval-runner.js +125 -125
  199. package/scripts/lib/metricas-dora.js +204 -204
  200. package/scripts/lib/notificaciones-telegram.js +1 -0
  201. package/scripts/lib/npm-version.js +1 -0
  202. package/scripts/lib/paquetes-conocidos.js +50 -50
  203. package/scripts/lib/plan-lock.js +61 -13
  204. package/scripts/lib/policy-broker.js +338 -0
  205. package/scripts/lib/policy-bundle.js +342 -0
  206. package/scripts/lib/policy-context-provider.js +310 -0
  207. package/scripts/lib/policy-contract.js +479 -0
  208. package/scripts/lib/policy-verifier-utils.js +65 -0
  209. package/scripts/lib/pr-analyzer.js +399 -399
  210. package/scripts/lib/principal-verifier.js +178 -0
  211. package/scripts/lib/prompt-builder.js +264 -264
  212. package/scripts/lib/resolver-plan-fase.js +37 -37
  213. package/scripts/lib/rrf-fusion.js +175 -175
  214. package/scripts/lib/runtime-adapter-contract.js +267 -0
  215. package/scripts/lib/runtime-artifact-verifier.js +426 -0
  216. package/scripts/lib/runtime-build-attestation.js +127 -0
  217. package/scripts/lib/runtime-bundle-installer.js +586 -0
  218. package/scripts/lib/runtime-compiler.js +327 -0
  219. package/scripts/lib/runtime-conformance.js +202 -0
  220. package/scripts/lib/runtime-doctor-core3.js +567 -0
  221. package/scripts/lib/runtime-doctor-input.js +59 -0
  222. package/scripts/lib/runtime-operation-adapter.js +267 -0
  223. package/scripts/lib/schema-version.js +164 -164
  224. package/scripts/lib/semantic-search.js +252 -252
  225. package/scripts/lib/signed-envelope.js +545 -0
  226. package/scripts/lib/single-use-store.js +359 -0
  227. package/scripts/lib/skills-externas.js +31 -0
  228. package/scripts/lib/transformadores/codex.js +15 -8
  229. package/scripts/lib/transformadores/gemini.js +79 -5
  230. package/scripts/lib/w4-attestation-adapter.js +158 -0
  231. package/scripts/lib/w4-canario.js +337 -0
  232. package/scripts/lib/w4-claims.js +182 -0
  233. package/scripts/lib/w4-corpus-generador.js +542 -0
  234. package/scripts/lib/w4-gate-c5.js +115 -0
  235. package/scripts/lib/w4-harness-bajo-prueba.js +155 -0
  236. package/scripts/lib/w4-matriz-combos.js +55 -0
  237. package/scripts/lib/w4-motor-mutacion.js +1348 -0
  238. package/scripts/lib/w4-motor-replay.js +735 -0
  239. package/scripts/lib/w4-pin-origen.js +54 -0
  240. package/scripts/lib/w4-publicar-request.js +132 -0
  241. package/scripts/lib/w4-revocacion.js +62 -0
  242. package/scripts/lib/w4-runtimes-core3.js +38 -0
  243. package/scripts/lib/w4-scorer-certificacion.js +692 -0
  244. package/scripts/lib/w4-superficie-candidato.js +49 -0
  245. package/scripts/lib/w4-veredicto.js +452 -0
  246. package/scripts/lib/w4-verificar-veredicto.js +302 -0
  247. package/scripts/limpiar-artefactos-python.js +131 -131
  248. package/scripts/migrar-csv-a-array.js +168 -168
  249. package/scripts/migrar-fase-dominio.js +200 -200
  250. package/scripts/migrar-gates-legacy.js +108 -0
  251. package/scripts/publicar-certification-request.js +115 -0
  252. package/scripts/runtime-doctor.js +107 -0
  253. package/scripts/tui/componentes/selector-multi.js +189 -189
  254. package/scripts/tui/componentes/selector-unico.js +158 -158
  255. package/scripts/tui/ejecutores.js +375 -375
  256. package/scripts/tui/lib/colores.js +129 -129
  257. package/scripts/tui/lib/render.js +264 -264
  258. package/scripts/tui/lib/teclas.js +113 -113
  259. package/scripts/tui/pantallas/install-wizard.js +12 -7
  260. package/scripts/tui/pantallas/menu-principal.js +52 -52
  261. package/scripts/tui/pantallas/progreso.js +274 -274
  262. package/scripts/tui/pantallas/resumen.js +132 -132
  263. package/scripts/validar-userland-vacio.js +110 -110
  264. package/scripts/verificar-aislamiento-swl-eval.js +87 -0
  265. package/scripts/verificar-empaquetado-downstream.js +375 -0
  266. package/scripts/verificar-loop-constructor.js +215 -0
  267. package/scripts/verificar-trazabilidad.js +13 -6
  268. package/scripts/verificar-veredicto-real.js +84 -0
@@ -1,171 +1,171 @@
1
- 'use strict';
2
-
3
- /**
4
- * eval-quality.js — Métricas de calidad para outputs estructurados de SWL.
5
- *
6
- * Patrón adoptado de `temp/agentmemory-main/src/eval/quality.ts`. Adaptado a
7
- * swl-ses: scoring de aprendizajes, instintos, y resultados de búsqueda
8
- * memoria.
9
- *
10
- * Cada función devuelve un score en [0, 100]. Los puntos se asignan por
11
- * "presencia y calidad de campos clave" — un output con todos los campos
12
- * tiene score 100, un output trivial tiene score bajo.
13
- *
14
- * Funciones puras zero-deps.
15
- *
16
- * @module scripts/lib/eval-quality
17
- */
18
-
19
- // ── helpers ───────────────────────────────────────────────────────────────────
20
-
21
- function clamp(n, min, max) {
22
- if (Number.isNaN(n)) return min;
23
- return Math.max(min, Math.min(max, n));
24
- }
25
-
26
- // ── scoring de outputs ────────────────────────────────────────────────────────
27
-
28
- /**
29
- * Score de calidad de una observación comprimida.
30
- * Adaptado de scoreCompression() de agentmemory.
31
- *
32
- * Distribución de puntos:
33
- * - facts presentes (no vacíos): 25
34
- * - facts ≥ 3: +10
35
- * - narrative ≥ 20 chars: 20
36
- * - narrative ≥ 50 chars: +5
37
- * - title 5-120 chars: 15
38
- * - concepts presentes: 15
39
- * - importance ∈ [1, 10]: 10
40
- *
41
- * @param {object} obs
42
- * @returns {number} en [0, 100]
43
- */
44
- function scoreObservacion(obs) {
45
- let score = 0;
46
- if (Array.isArray(obs?.facts) && obs.facts.length > 0) score += 25;
47
- if (Array.isArray(obs?.facts) && obs.facts.length >= 3) score += 10;
48
- if (typeof obs?.narrative === 'string' && obs.narrative.length >= 20) score += 20;
49
- if (typeof obs?.narrative === 'string' && obs.narrative.length >= 50) score += 5;
50
- if (typeof obs?.title === 'string' && obs.title.length >= 5 && obs.title.length <= 120) score += 15;
51
- if (Array.isArray(obs?.concepts) && obs.concepts.length > 0) score += 15;
52
- if (typeof obs?.importance === 'number' && obs.importance >= 1 && obs.importance <= 10) score += 10;
53
- return clamp(score, 0, 100);
54
- }
55
-
56
- /**
57
- * Score de calidad de un resumen de sesión.
58
- * Adaptado de scoreSummary() de agentmemory.
59
- *
60
- * @param {object} summary
61
- * @returns {number} en [0, 100]
62
- */
63
- function scoreResumen(summary) {
64
- let score = 0;
65
- if (typeof summary?.title === 'string' && summary.title.length >= 5) score += 20;
66
- if (typeof summary?.narrative === 'string' && summary.narrative.length >= 20) score += 25;
67
- if (typeof summary?.narrative === 'string' && summary.narrative.length >= 100) score += 5;
68
- if (Array.isArray(summary?.keyDecisions) && summary.keyDecisions.length > 0) score += 20;
69
- if (Array.isArray(summary?.filesModified) && summary.filesModified.length > 0) score += 15;
70
- if (Array.isArray(summary?.concepts) && summary.concepts.length > 0) score += 15;
71
- return clamp(score, 0, 100);
72
- }
73
-
74
- /**
75
- * Score de relevancia de un contexto inyectado.
76
- * Adaptado de scoreContextRelevance() de agentmemory.
77
- *
78
- * @param {string} context - Texto del contexto.
79
- * @param {string} project - Nombre del proyecto.
80
- * @returns {number} en [0, 100]
81
- */
82
- function scoreRelevanciaContexto(context, project) {
83
- if (typeof context !== 'string') return 0;
84
- let score = 0;
85
- if (context.length > 0) score += 20;
86
- if (project && context.toLowerCase().includes(String(project).toLowerCase())) score += 20;
87
- if (context.includes('<')) score += 15;
88
- const sectionCount = (context.match(/<\w+>/g) || []).length;
89
- if (sectionCount >= 2) score += 15;
90
- if (sectionCount >= 4) score += 10;
91
- if (context.length >= 100) score += 10;
92
- if (context.length >= 500) score += 10;
93
- return clamp(score, 0, 100);
94
- }
95
-
96
- /**
97
- * Score de calidad de un aprendizaje SWL extraído de una sesión.
98
- * Específico de swl-ses (no en agentmemory).
99
- *
100
- * Distribución:
101
- * - título no vacío y descriptivo (≥ 10 chars): 20
102
- * - título empieza con fecha [YYYY-MM-DD]: 10
103
- * - cuerpo ≥ 100 chars: 20
104
- * - cuerpo ≥ 300 chars: +10
105
- * - tipo identificado (decisión|patrón|...): 15
106
- * - menciona archivo o regla concreta: 15
107
- * - tiene "trigger" o "criterio de disparo": 10
108
- *
109
- * @param {object} aprendizaje { titulo, contenido, tipo? }
110
- * @returns {number}
111
- */
112
- function scoreAprendizaje(aprendizaje) {
113
- let score = 0;
114
- const titulo = String(aprendizaje?.titulo || '');
115
- const contenido = String(aprendizaje?.contenido || '');
116
-
117
- if (titulo.length >= 10) score += 20;
118
- if (/^\[\d{4}-\d{2}-\d{2}\]/.test(titulo)) score += 10;
119
- if (contenido.length >= 100) score += 20;
120
- if (contenido.length >= 300) score += 10;
121
-
122
- const tipos = ['decisión', 'patrón', 'anti-patrón', 'bug-fix', 'descubrimiento', 'gotcha'];
123
- if (aprendizaje?.tipo && tipos.some(t => String(aprendizaje.tipo).includes(t))) {
124
- score += 15;
125
- }
126
-
127
- if (/`[^`]+\.(js|ts|md|py|json|yaml)`|`[^`]+\/[^`]+`/.test(contenido)) score += 15;
128
- if (/trigger|criterio de disparo|cuando .+ entonces/i.test(contenido)) score += 10;
129
-
130
- return clamp(score, 0, 100);
131
- }
132
-
133
- /**
134
- * Score de calidad de un instinto.
135
- * Específico de swl-ses.
136
- *
137
- * Distribución:
138
- * - pattern presente: 25
139
- * - pattern ≥ 30 chars: +10
140
- * - confidence ∈ [0, 1]: 15
141
- * - status válido (active|degraded|archived): 10
142
- * - source_sessions o source_agents declarado: 15
143
- * - evidence_count ≥ 1: 15
144
- * - last_validated_at presente: 10
145
- *
146
- * @param {object} instinto
147
- * @returns {number}
148
- */
149
- function scoreInstinto(instinto) {
150
- let score = 0;
151
- const pattern = String(instinto?.pattern || '');
152
- if (pattern.length > 0) score += 25;
153
- if (pattern.length >= 30) score += 10;
154
- if (typeof instinto?.confidence === 'number' && instinto.confidence >= 0 && instinto.confidence <= 1) score += 15;
155
- if (['active', 'degraded', 'archived'].includes(instinto?.status)) score += 10;
156
- if ((Array.isArray(instinto?.source_sessions) && instinto.source_sessions.length > 0) ||
157
- (Array.isArray(instinto?.source_agents) && instinto.source_agents.length > 0)) score += 15;
158
- if (typeof instinto?.evidence_count === 'number' && instinto.evidence_count >= 1) score += 15;
159
- if (instinto?.last_validated_at || instinto?.last_validated) score += 10;
160
- return clamp(score, 0, 100);
161
- }
162
-
163
- // ── exports ───────────────────────────────────────────────────────────────────
164
-
165
- module.exports = {
166
- scoreObservacion,
167
- scoreResumen,
168
- scoreRelevanciaContexto,
169
- scoreAprendizaje,
170
- scoreInstinto,
171
- };
1
+ 'use strict';
2
+
3
+ /**
4
+ * eval-quality.js — Métricas de calidad para outputs estructurados de SWL.
5
+ *
6
+ * Patrón adoptado de `temp/agentmemory-main/src/eval/quality.ts`. Adaptado a
7
+ * swl-ses: scoring de aprendizajes, instintos, y resultados de búsqueda
8
+ * memoria.
9
+ *
10
+ * Cada función devuelve un score en [0, 100]. Los puntos se asignan por
11
+ * "presencia y calidad de campos clave" — un output con todos los campos
12
+ * tiene score 100, un output trivial tiene score bajo.
13
+ *
14
+ * Funciones puras zero-deps.
15
+ *
16
+ * @module scripts/lib/eval-quality
17
+ */
18
+
19
+ // ── helpers ───────────────────────────────────────────────────────────────────
20
+
21
+ function clamp(n, min, max) {
22
+ if (Number.isNaN(n)) return min;
23
+ return Math.max(min, Math.min(max, n));
24
+ }
25
+
26
+ // ── scoring de outputs ────────────────────────────────────────────────────────
27
+
28
+ /**
29
+ * Score de calidad de una observación comprimida.
30
+ * Adaptado de scoreCompression() de agentmemory.
31
+ *
32
+ * Distribución de puntos:
33
+ * - facts presentes (no vacíos): 25
34
+ * - facts ≥ 3: +10
35
+ * - narrative ≥ 20 chars: 20
36
+ * - narrative ≥ 50 chars: +5
37
+ * - title 5-120 chars: 15
38
+ * - concepts presentes: 15
39
+ * - importance ∈ [1, 10]: 10
40
+ *
41
+ * @param {object} obs
42
+ * @returns {number} en [0, 100]
43
+ */
44
+ function scoreObservacion(obs) {
45
+ let score = 0;
46
+ if (Array.isArray(obs?.facts) && obs.facts.length > 0) score += 25;
47
+ if (Array.isArray(obs?.facts) && obs.facts.length >= 3) score += 10;
48
+ if (typeof obs?.narrative === 'string' && obs.narrative.length >= 20) score += 20;
49
+ if (typeof obs?.narrative === 'string' && obs.narrative.length >= 50) score += 5;
50
+ if (typeof obs?.title === 'string' && obs.title.length >= 5 && obs.title.length <= 120) score += 15;
51
+ if (Array.isArray(obs?.concepts) && obs.concepts.length > 0) score += 15;
52
+ if (typeof obs?.importance === 'number' && obs.importance >= 1 && obs.importance <= 10) score += 10;
53
+ return clamp(score, 0, 100);
54
+ }
55
+
56
+ /**
57
+ * Score de calidad de un resumen de sesión.
58
+ * Adaptado de scoreSummary() de agentmemory.
59
+ *
60
+ * @param {object} summary
61
+ * @returns {number} en [0, 100]
62
+ */
63
+ function scoreResumen(summary) {
64
+ let score = 0;
65
+ if (typeof summary?.title === 'string' && summary.title.length >= 5) score += 20;
66
+ if (typeof summary?.narrative === 'string' && summary.narrative.length >= 20) score += 25;
67
+ if (typeof summary?.narrative === 'string' && summary.narrative.length >= 100) score += 5;
68
+ if (Array.isArray(summary?.keyDecisions) && summary.keyDecisions.length > 0) score += 20;
69
+ if (Array.isArray(summary?.filesModified) && summary.filesModified.length > 0) score += 15;
70
+ if (Array.isArray(summary?.concepts) && summary.concepts.length > 0) score += 15;
71
+ return clamp(score, 0, 100);
72
+ }
73
+
74
+ /**
75
+ * Score de relevancia de un contexto inyectado.
76
+ * Adaptado de scoreContextRelevance() de agentmemory.
77
+ *
78
+ * @param {string} context - Texto del contexto.
79
+ * @param {string} project - Nombre del proyecto.
80
+ * @returns {number} en [0, 100]
81
+ */
82
+ function scoreRelevanciaContexto(context, project) {
83
+ if (typeof context !== 'string') return 0;
84
+ let score = 0;
85
+ if (context.length > 0) score += 20;
86
+ if (project && context.toLowerCase().includes(String(project).toLowerCase())) score += 20;
87
+ if (context.includes('<')) score += 15;
88
+ const sectionCount = (context.match(/<\w+>/g) || []).length;
89
+ if (sectionCount >= 2) score += 15;
90
+ if (sectionCount >= 4) score += 10;
91
+ if (context.length >= 100) score += 10;
92
+ if (context.length >= 500) score += 10;
93
+ return clamp(score, 0, 100);
94
+ }
95
+
96
+ /**
97
+ * Score de calidad de un aprendizaje SWL extraído de una sesión.
98
+ * Específico de swl-ses (no en agentmemory).
99
+ *
100
+ * Distribución:
101
+ * - título no vacío y descriptivo (≥ 10 chars): 20
102
+ * - título empieza con fecha [YYYY-MM-DD]: 10
103
+ * - cuerpo ≥ 100 chars: 20
104
+ * - cuerpo ≥ 300 chars: +10
105
+ * - tipo identificado (decisión|patrón|...): 15
106
+ * - menciona archivo o regla concreta: 15
107
+ * - tiene "trigger" o "criterio de disparo": 10
108
+ *
109
+ * @param {object} aprendizaje { titulo, contenido, tipo? }
110
+ * @returns {number}
111
+ */
112
+ function scoreAprendizaje(aprendizaje) {
113
+ let score = 0;
114
+ const titulo = String(aprendizaje?.titulo || '');
115
+ const contenido = String(aprendizaje?.contenido || '');
116
+
117
+ if (titulo.length >= 10) score += 20;
118
+ if (/^\[\d{4}-\d{2}-\d{2}\]/.test(titulo)) score += 10;
119
+ if (contenido.length >= 100) score += 20;
120
+ if (contenido.length >= 300) score += 10;
121
+
122
+ const tipos = ['decisión', 'patrón', 'anti-patrón', 'bug-fix', 'descubrimiento', 'gotcha'];
123
+ if (aprendizaje?.tipo && tipos.some(t => String(aprendizaje.tipo).includes(t))) {
124
+ score += 15;
125
+ }
126
+
127
+ if (/`[^`]+\.(js|ts|md|py|json|yaml)`|`[^`]+\/[^`]+`/.test(contenido)) score += 15;
128
+ if (/trigger|criterio de disparo|cuando .+ entonces/i.test(contenido)) score += 10;
129
+
130
+ return clamp(score, 0, 100);
131
+ }
132
+
133
+ /**
134
+ * Score de calidad de un instinto.
135
+ * Específico de swl-ses.
136
+ *
137
+ * Distribución:
138
+ * - pattern presente: 25
139
+ * - pattern ≥ 30 chars: +10
140
+ * - confidence ∈ [0, 1]: 15
141
+ * - status válido (active|degraded|archived): 10
142
+ * - source_sessions o source_agents declarado: 15
143
+ * - evidence_count ≥ 1: 15
144
+ * - last_validated_at presente: 10
145
+ *
146
+ * @param {object} instinto
147
+ * @returns {number}
148
+ */
149
+ function scoreInstinto(instinto) {
150
+ let score = 0;
151
+ const pattern = String(instinto?.pattern || '');
152
+ if (pattern.length > 0) score += 25;
153
+ if (pattern.length >= 30) score += 10;
154
+ if (typeof instinto?.confidence === 'number' && instinto.confidence >= 0 && instinto.confidence <= 1) score += 15;
155
+ if (['active', 'degraded', 'archived'].includes(instinto?.status)) score += 10;
156
+ if ((Array.isArray(instinto?.source_sessions) && instinto.source_sessions.length > 0) ||
157
+ (Array.isArray(instinto?.source_agents) && instinto.source_agents.length > 0)) score += 15;
158
+ if (typeof instinto?.evidence_count === 'number' && instinto.evidence_count >= 1) score += 15;
159
+ if (instinto?.last_validated_at || instinto?.last_validated) score += 10;
160
+ return clamp(score, 0, 100);
161
+ }
162
+
163
+ // ── exports ───────────────────────────────────────────────────────────────────
164
+
165
+ module.exports = {
166
+ scoreObservacion,
167
+ scoreResumen,
168
+ scoreRelevanciaContexto,
169
+ scoreAprendizaje,
170
+ scoreInstinto,
171
+ };
@@ -1,144 +1,144 @@
1
- 'use strict';
2
-
3
- /**
4
- * eval-schemas.js — Schemas JSON-lite para evaluación de outputs SWL.
5
- *
6
- * Patrón adoptado de `temp/agentmemory-main/src/eval/schemas.ts`. Adaptado a
7
- * swl-ses: sin Zod (sería dep externa). Uso JSON Schema-lite con validador
8
- * propio en `eval-validator.js`. Funciones puras zero-deps.
9
- *
10
- * Cada schema describe la estructura esperada de un output evaluable. El
11
- * validador devuelve `{ valid, errors[] }` para que el caller decida.
12
- *
13
- * @module scripts/lib/eval-schemas
14
- */
15
-
16
- // ── tipos enumerados ──────────────────────────────────────────────────────────
17
-
18
- const TIPOS_OBSERVACION = [
19
- 'file_read', 'file_write', 'file_edit',
20
- 'command_run', 'search', 'web_fetch',
21
- 'conversation', 'error', 'decision',
22
- 'discovery', 'subagent', 'notification',
23
- 'task', 'other',
24
- ];
25
-
26
- const TIPOS_MEMORIA = [
27
- 'pattern', 'preference', 'architecture',
28
- 'bug', 'workflow', 'fact',
29
- ];
30
-
31
- const TIPOS_RELACION = [
32
- 'supersedes', 'extends', 'derives', 'contradicts', 'related',
33
- ];
34
-
35
- // ── schemas JSON-lite ─────────────────────────────────────────────────────────
36
-
37
- /**
38
- * Schema para output de compresión de observación.
39
- * Estructura compatible con `CompressOutputSchema` de agentmemory.
40
- */
41
- const COMPRESS_OUTPUT_SCHEMA = {
42
- type: 'object',
43
- required: ['type', 'title', 'facts', 'narrative', 'concepts', 'files', 'importance'],
44
- properties: {
45
- type: { type: 'string', enum: TIPOS_OBSERVACION },
46
- title: { type: 'string', minLength: 1, maxLength: 120 },
47
- subtitle: { type: 'string' },
48
- facts: { type: 'array', minItems: 1, items: { type: 'string' } },
49
- narrative: { type: 'string', minLength: 10 },
50
- concepts: { type: 'array', items: { type: 'string' } },
51
- files: { type: 'array', items: { type: 'string' } },
52
- importance: { type: 'integer', minimum: 1, maximum: 10 },
53
- },
54
- };
55
-
56
- /**
57
- * Schema para output de resumen de sesión.
58
- */
59
- const SUMMARY_OUTPUT_SCHEMA = {
60
- type: 'object',
61
- required: ['title', 'narrative', 'keyDecisions', 'filesModified', 'concepts'],
62
- properties: {
63
- title: { type: 'string', minLength: 1 },
64
- narrative: { type: 'string', minLength: 20 },
65
- keyDecisions: { type: 'array', items: { type: 'string' } },
66
- filesModified: { type: 'array', items: { type: 'string' } },
67
- concepts: { type: 'array', items: { type: 'string' } },
68
- },
69
- };
70
-
71
- /**
72
- * Schema para input de búsqueda.
73
- */
74
- const SEARCH_INPUT_SCHEMA = {
75
- type: 'object',
76
- required: ['query'],
77
- properties: {
78
- query: { type: 'string', minLength: 1 },
79
- limit: { type: 'integer', minimum: 1 },
80
- },
81
- };
82
-
83
- /**
84
- * Schema para input de "remember" (guardar memoria).
85
- */
86
- const REMEMBER_INPUT_SCHEMA = {
87
- type: 'object',
88
- required: ['content'],
89
- properties: {
90
- content: { type: 'string', minLength: 1 },
91
- type: { type: 'string', enum: TIPOS_MEMORIA },
92
- concepts: { type: 'array', items: { type: 'string' } },
93
- files: { type: 'array', items: { type: 'string' } },
94
- },
95
- };
96
-
97
- /**
98
- * Schema para resultado de evaluación.
99
- */
100
- const EVAL_RESULT_SCHEMA = {
101
- type: 'object',
102
- required: ['valid', 'qualityScore', 'latencyMs', 'functionId'],
103
- properties: {
104
- valid: { type: 'boolean' },
105
- errors: { type: 'array', items: { type: 'string' } },
106
- qualityScore: { type: 'number', minimum: 0, maximum: 100 },
107
- latencyMs: { type: 'number', minimum: 0 },
108
- functionId: { type: 'string', minLength: 1 },
109
- metadata: { type: 'object' },
110
- },
111
- };
112
-
113
- /**
114
- * Schema para resultado de búsqueda en memoria SWL.
115
- */
116
- const MEMORY_SEARCH_RESULT_SCHEMA = {
117
- type: 'object',
118
- required: ['id', 'tipo', 'titulo', 'fecha', 'relevancia'],
119
- properties: {
120
- id: { type: 'string', minLength: 1 },
121
- tipo: { type: 'string', enum: ['aprendizaje', 'sesion', 'instinto'] },
122
- titulo: { type: 'string' },
123
- fecha: { type: 'string' },
124
- relevancia: { type: 'number', minimum: 0, maximum: 1 },
125
- combinedScore: { type: 'number', minimum: 0 },
126
- confidence: { type: 'number', minimum: 0, maximum: 1 },
127
- },
128
- };
129
-
130
- // ── exports ───────────────────────────────────────────────────────────────────
131
-
132
- module.exports = {
133
- // Schemas
134
- COMPRESS_OUTPUT_SCHEMA,
135
- SUMMARY_OUTPUT_SCHEMA,
136
- SEARCH_INPUT_SCHEMA,
137
- REMEMBER_INPUT_SCHEMA,
138
- EVAL_RESULT_SCHEMA,
139
- MEMORY_SEARCH_RESULT_SCHEMA,
140
- // Enums
141
- TIPOS_OBSERVACION,
142
- TIPOS_MEMORIA,
143
- TIPOS_RELACION,
144
- };
1
+ 'use strict';
2
+
3
+ /**
4
+ * eval-schemas.js — Schemas JSON-lite para evaluación de outputs SWL.
5
+ *
6
+ * Patrón adoptado de `temp/agentmemory-main/src/eval/schemas.ts`. Adaptado a
7
+ * swl-ses: sin Zod (sería dep externa). Uso JSON Schema-lite con validador
8
+ * propio en `eval-validator.js`. Funciones puras zero-deps.
9
+ *
10
+ * Cada schema describe la estructura esperada de un output evaluable. El
11
+ * validador devuelve `{ valid, errors[] }` para que el caller decida.
12
+ *
13
+ * @module scripts/lib/eval-schemas
14
+ */
15
+
16
+ // ── tipos enumerados ──────────────────────────────────────────────────────────
17
+
18
+ const TIPOS_OBSERVACION = [
19
+ 'file_read', 'file_write', 'file_edit',
20
+ 'command_run', 'search', 'web_fetch',
21
+ 'conversation', 'error', 'decision',
22
+ 'discovery', 'subagent', 'notification',
23
+ 'task', 'other',
24
+ ];
25
+
26
+ const TIPOS_MEMORIA = [
27
+ 'pattern', 'preference', 'architecture',
28
+ 'bug', 'workflow', 'fact',
29
+ ];
30
+
31
+ const TIPOS_RELACION = [
32
+ 'supersedes', 'extends', 'derives', 'contradicts', 'related',
33
+ ];
34
+
35
+ // ── schemas JSON-lite ─────────────────────────────────────────────────────────
36
+
37
+ /**
38
+ * Schema para output de compresión de observación.
39
+ * Estructura compatible con `CompressOutputSchema` de agentmemory.
40
+ */
41
+ const COMPRESS_OUTPUT_SCHEMA = {
42
+ type: 'object',
43
+ required: ['type', 'title', 'facts', 'narrative', 'concepts', 'files', 'importance'],
44
+ properties: {
45
+ type: { type: 'string', enum: TIPOS_OBSERVACION },
46
+ title: { type: 'string', minLength: 1, maxLength: 120 },
47
+ subtitle: { type: 'string' },
48
+ facts: { type: 'array', minItems: 1, items: { type: 'string' } },
49
+ narrative: { type: 'string', minLength: 10 },
50
+ concepts: { type: 'array', items: { type: 'string' } },
51
+ files: { type: 'array', items: { type: 'string' } },
52
+ importance: { type: 'integer', minimum: 1, maximum: 10 },
53
+ },
54
+ };
55
+
56
+ /**
57
+ * Schema para output de resumen de sesión.
58
+ */
59
+ const SUMMARY_OUTPUT_SCHEMA = {
60
+ type: 'object',
61
+ required: ['title', 'narrative', 'keyDecisions', 'filesModified', 'concepts'],
62
+ properties: {
63
+ title: { type: 'string', minLength: 1 },
64
+ narrative: { type: 'string', minLength: 20 },
65
+ keyDecisions: { type: 'array', items: { type: 'string' } },
66
+ filesModified: { type: 'array', items: { type: 'string' } },
67
+ concepts: { type: 'array', items: { type: 'string' } },
68
+ },
69
+ };
70
+
71
+ /**
72
+ * Schema para input de búsqueda.
73
+ */
74
+ const SEARCH_INPUT_SCHEMA = {
75
+ type: 'object',
76
+ required: ['query'],
77
+ properties: {
78
+ query: { type: 'string', minLength: 1 },
79
+ limit: { type: 'integer', minimum: 1 },
80
+ },
81
+ };
82
+
83
+ /**
84
+ * Schema para input de "remember" (guardar memoria).
85
+ */
86
+ const REMEMBER_INPUT_SCHEMA = {
87
+ type: 'object',
88
+ required: ['content'],
89
+ properties: {
90
+ content: { type: 'string', minLength: 1 },
91
+ type: { type: 'string', enum: TIPOS_MEMORIA },
92
+ concepts: { type: 'array', items: { type: 'string' } },
93
+ files: { type: 'array', items: { type: 'string' } },
94
+ },
95
+ };
96
+
97
+ /**
98
+ * Schema para resultado de evaluación.
99
+ */
100
+ const EVAL_RESULT_SCHEMA = {
101
+ type: 'object',
102
+ required: ['valid', 'qualityScore', 'latencyMs', 'functionId'],
103
+ properties: {
104
+ valid: { type: 'boolean' },
105
+ errors: { type: 'array', items: { type: 'string' } },
106
+ qualityScore: { type: 'number', minimum: 0, maximum: 100 },
107
+ latencyMs: { type: 'number', minimum: 0 },
108
+ functionId: { type: 'string', minLength: 1 },
109
+ metadata: { type: 'object' },
110
+ },
111
+ };
112
+
113
+ /**
114
+ * Schema para resultado de búsqueda en memoria SWL.
115
+ */
116
+ const MEMORY_SEARCH_RESULT_SCHEMA = {
117
+ type: 'object',
118
+ required: ['id', 'tipo', 'titulo', 'fecha', 'relevancia'],
119
+ properties: {
120
+ id: { type: 'string', minLength: 1 },
121
+ tipo: { type: 'string', enum: ['aprendizaje', 'sesion', 'instinto'] },
122
+ titulo: { type: 'string' },
123
+ fecha: { type: 'string' },
124
+ relevancia: { type: 'number', minimum: 0, maximum: 1 },
125
+ combinedScore: { type: 'number', minimum: 0 },
126
+ confidence: { type: 'number', minimum: 0, maximum: 1 },
127
+ },
128
+ };
129
+
130
+ // ── exports ───────────────────────────────────────────────────────────────────
131
+
132
+ module.exports = {
133
+ // Schemas
134
+ COMPRESS_OUTPUT_SCHEMA,
135
+ SUMMARY_OUTPUT_SCHEMA,
136
+ SEARCH_INPUT_SCHEMA,
137
+ REMEMBER_INPUT_SCHEMA,
138
+ EVAL_RESULT_SCHEMA,
139
+ MEMORY_SEARCH_RESULT_SCHEMA,
140
+ // Enums
141
+ TIPOS_OBSERVACION,
142
+ TIPOS_MEMORIA,
143
+ TIPOS_RELACION,
144
+ };