n8n-nodes-tembory 1.0.12 → 1.0.14

package/README.md

@@ -2,7 +2,15 @@
 
 Node de memoria operacional da Tembory para agentes de IA no n8n.
 
-Versao atual: `1.0.12`.
+Versao atual: `1.0.14`.
+
+## 1.0.14
+
+- Adiciona Active Summary persistente por namespace/thread para continuar conversas depois de minutos, horas ou dias.
+- Separa cache tecnico curto do resumo SLM da memoria ativa persistente.
+- Aumenta os defaults do resumo SLM nos presets de producao para preservar mais inteligencia antes de comprimir para o agente.
+- Adiciona diagnosticos para saber se o contexto veio de SLM novo, active summary ou cache tecnico.
+- Expõe controles avancados de Active Summary e cache tecnico no node.
 
 O Tembory entrega contexto rico para o AI Agent sem depender apenas do historico textual da conversa. Ele combina memoria semantica, working memory, decision state, fatos estaveis do lead, historico de tools, estado operacional, action ledger, timeline de entidades, compressao de memoria, grafo, mensagens recentes e diagnosticos.
 
@@ -10,6 +18,8 @@ Nos presets de producao, o contexto e organizado como um pacote acionavel: secoe
 
 Quando um modelo barato esta conectado ao Tembory, os presets de producao usam esse SLM para gerar um resumo ativo do contexto. Em modo `auto`, ele organiza tanto o estado atual (`working_memory`, `tool_history`, `operational_state`, `decision_state`) quanto memorias vetoriais recuperadas. Assim o primeiro turno ja pode receber highlights uteis mesmo antes de haver resultados do banco vetorial.
 
+O resumo do SLM usa limites conservadores e cache curto por input para evitar chamadas duplicadas quando o AI Agent carrega a memoria mais de uma vez no mesmo turno.
+
 ## Smoke tecnico
 
 Antes de publicar uma versao, rode:

package/dist/nodes/Mem0/Mem0Memory.node.js

@@ -862,13 +862,17 @@ const getMemoryStore = (ctx) => {
         data.tembory.workingMemory = data.tembory.workingMemory || {};
         data.tembory.decisionState = data.tembory.decisionState || {};
         data.tembory.memoryCompression = data.tembory.memoryCompression || {};
+        data.tembory.activeSummary = data.tembory.activeSummary || {};
+        data.tembory.connectedModelSummaryCache = data.tembory.connectedModelSummaryCache || {};
         return data.tembory;
     }
     catch {
-        global.__temboryMemory = global.__temboryMemory || { toolHistory: {}, recentMessages: {}, profileFacts: {}, workingMemory: {}, decisionState: {}, memoryCompression: {} };
+        global.__temboryMemory = global.__temboryMemory || { toolHistory: {}, recentMessages: {}, profileFacts: {}, workingMemory: {}, decisionState: {}, memoryCompression: {}, activeSummary: {}, connectedModelSummaryCache: {} };
         global.__temboryMemory.workingMemory = global.__temboryMemory.workingMemory || {};
         global.__temboryMemory.decisionState = global.__temboryMemory.decisionState || {};
         global.__temboryMemory.memoryCompression = global.__temboryMemory.memoryCompression || {};
+        global.__temboryMemory.activeSummary = global.__temboryMemory.activeSummary || {};
+        global.__temboryMemory.connectedModelSummaryCache = global.__temboryMemory.connectedModelSummaryCache || {};
         return global.__temboryMemory;
     }
 };
@@ -910,6 +914,12 @@ const applyOperationalPreset = (advanced = {}) => {
         productionBalanced: {
             summarySource: 'auto',
             includeConnectedModelSummary: true,
+            includeActiveSummary: true,
+            persistActiveSummary: true,
+            activeSummaryMaxChars: 1800,
+            activeSummaryRetentionDays: 30,
+            enableTransientSummaryCache: true,
+            transientSummaryCacheTTLSeconds: 300,
             compactStateSections: true,
             includeContextHeader: true,
             includeSummary: true,
@@ -935,12 +945,18 @@ const applyOperationalPreset = (advanced = {}) => {
             recentMessagesLastN: 6,
             vectorMemoryMaxChars: 360,
             contextMaxChars: 10000,
-            connectedModelSummaryMaxChars: 900,
-            connectedModelSummaryInputMaxChars: 3200,
+            connectedModelSummaryMaxChars: 1200,
+            connectedModelSummaryInputMaxChars: 4200,
         },
         productionCheap: {
             summarySource: 'activeContext',
             includeConnectedModelSummary: true,
+            includeActiveSummary: true,
+            persistActiveSummary: true,
+            activeSummaryMaxChars: 1500,
+            activeSummaryRetentionDays: 30,
+            enableTransientSummaryCache: true,
+            transientSummaryCacheTTLSeconds: 300,
             compactStateSections: true,
             includeContextHeader: true,
             includeSummary: false,
@@ -966,12 +982,18 @@ const applyOperationalPreset = (advanced = {}) => {
             recentMessagesLastN: 4,
             vectorMemoryMaxChars: 260,
             contextMaxChars: 7000,
-            connectedModelSummaryMaxChars: 700,
-            connectedModelSummaryInputMaxChars: 2400,
+            connectedModelSummaryMaxChars: 1000,
+            connectedModelSummaryInputMaxChars: 3600,
         },
         productionNano: {
             summarySource: 'auto',
             includeConnectedModelSummary: true,
+            includeActiveSummary: true,
+            persistActiveSummary: true,
+            activeSummaryMaxChars: 1600,
+            activeSummaryRetentionDays: 30,
+            enableTransientSummaryCache: true,
+            transientSummaryCacheTTLSeconds: 300,
             compactForAgent: true,
             includeContextHeader: true,
             includeSummary: true,
@@ -997,9 +1019,9 @@ const applyOperationalPreset = (advanced = {}) => {
             toolHistoryLastN: 6,
             recentMessagesLastN: 2,
             vectorMemoryMaxChars: 220,
-            contextMaxChars: 6000,
-            connectedModelSummaryMaxChars: 700,
-            connectedModelSummaryInputMaxChars: 2200,
+            contextMaxChars: 7000,
+            connectedModelSummaryMaxChars: 1200,
+            connectedModelSummaryInputMaxChars: 4200,
         },
         audit: {
             summarySource: 'auto',
@@ -1580,7 +1602,34 @@ const cleanModelSummaryText = (value, max = 900) => {
         .trim();
     return truncate(text, max);
 };
-const buildConnectedModelSummaryInput = ({ query, profileFacts, workingMemory, decisionState, memoryCompression, operationalState, toolHistory, recentMessages, vectorMemories, highlights, adv }) => {
+const activeSummaryIsFresh = (entry, adv = {}) => {
+    if (!entry || !entry.summary)
+        return false;
+    const retentionDays = Number(adv.activeSummaryRetentionDays ?? 30);
+    if (retentionDays <= 0)
+        return true;
+    return Date.now() - Number(entry.updatedAt || entry.at || 0) < retentionDays * 86400000;
+};
+const readActiveSummary = (store, key, adv = {}) => {
+    if (adv.includeActiveSummary === false)
+        return '';
+    const entry = store.activeSummary?.[key];
+    if (!activeSummaryIsFresh(entry, adv))
+        return '';
+    return truncate(String(entry.summary || ''), Number(adv.activeSummaryMaxChars || 1600));
+};
+const writeActiveSummary = (store, key, summary, adv = {}) => {
+    if (!store || !key || adv.persistActiveSummary === false || !summary)
+        return false;
+    store.activeSummary = store.activeSummary || {};
+    store.activeSummary[key] = {
+        summary: truncate(String(summary), Number(adv.activeSummaryMaxChars || 1600)),
+        updatedAt: Date.now(),
+        source: 'slm',
+    };
+    return true;
+};
+const buildConnectedModelSummaryInput = ({ query, activeSummary, profileFacts, workingMemory, decisionState, memoryCompression, operationalState, toolHistory, recentMessages, vectorMemories, highlights, adv }) => {
     const source = String(adv.summarySource || 'auto');
     if (source === 'off' || source === 'disabled' || adv.includeConnectedModelSummary === false)
         return '';
@@ -1588,6 +1637,7 @@ const buildConnectedModelSummaryInput = ({ query, profileFacts, workingMemory, d
     const includeActive = source === 'auto' || source === 'activeContext' || source === 'active';
     const payload = cleanContextValue({
         query: String(query || ''),
+        existing_active_summary: activeSummary ? truncate(activeSummary, Number(adv.activeSummaryMaxChars || 1600)) : undefined,
         active_context: includeActive ? {
             profile_facts: renderProfileFacts(profileFacts),
             working_memory: compactWorkingMemoryForAgent(workingMemory || {}),
@@ -1608,7 +1658,7 @@ const buildConnectedModelSummaryInput = ({ query, profileFacts, workingMemory, d
     const hasVectorSignal = includeVectors && Array.isArray(payload.vector_memories) && payload.vector_memories.length > 0;
     if (!hasActiveSignal && !hasVectorSignal)
         return '';
-    return truncate(safeStringify(payload), Number(adv.connectedModelSummaryInputMaxChars || 3000));
+    return truncate(safeStringify(payload), Number(adv.connectedModelSummaryInputMaxChars || 2000));
 };
 const invokeConnectedModelSummary = async (connectedLanguageModel, summaryInput, adv = {}) => {
     if (!connectedLanguageModel || typeof connectedLanguageModel.invoke !== 'function' || !summaryInput)
@@ -1616,10 +1666,10 @@ const invokeConnectedModelSummary = async (connectedLanguageModel, summaryInput,
     const response = await connectedLanguageModel.invoke([
         toBaseMessage({
             role: 'user',
-            content: `Organize the Tembory memory context for the next agent turn. Return concise Portuguese bullets only. Preserve IDs, dates, tool names, next action, do-not-repeat instructions, and contradictions. Do not invent facts.\n\nContext:\n${summaryInput}`,
+                            content: `Update the Tembory active summary for the next agent turn. Return only concise Portuguese bullets, no JSON and no markdown table. Preserve IDs, dates, tool names, confirmed decisions, pending actions, constraints, contradictions, and do-not-repeat instructions. Prefer durable useful context over raw logs. Do not invent facts.\n\nContext:\n${summaryInput}`,
         }),
     ]);
-    return cleanModelSummaryText(response, Number(adv.connectedModelSummaryMaxChars || 900));
+    return cleanModelSummaryText(response, Number(adv.connectedModelSummaryMaxChars || 1200));
 };
 const contextSizeOfMessages = (messages = []) => {
     const perMessage = (messages || []).map((message, index) => {
@@ -1651,6 +1701,7 @@ const wrapTemboryMemory = (memory, ctx, memoryKey) => new Proxy(memory, {
                                     context: response.temboryContext,
                                     contextText: response.temboryContextText,
                                     summary: response.temborySummary,
+                                    activeSummary: response.temboryActiveSummary,
                                     connectedModelSummary: response.temboryConnectedModelSummary,
                                     workingMemory: response.temboryWorkingMemory,
                                     decisionState: response.temboryDecisionState,
@@ -1698,7 +1749,7 @@ const wrapTemboryMemory = (memory, ctx, memoryKey) => new Proxy(memory, {
         return target[prop];
     },
 });
-const buildContextMessages = ({ payloadFormat, query, userId, profileFacts, workingMemory, decisionState, memoryCompression, operationalState, actionLedger, entityTimeline, vectorMemories, recentMessages, toolHistory, highlights, graph, diagnostics, connectedModelSummary, adv }) => {
+const buildContextMessages = ({ payloadFormat, query, userId, profileFacts, workingMemory, decisionState, memoryCompression, operationalState, actionLedger, entityTimeline, vectorMemories, recentMessages, toolHistory, highlights, graph, diagnostics, activeSummary, connectedModelSummary, adv }) => {
     const includeHeader = adv.includeContextHeader !== false;
     const includeSummary = adv.includeSummary !== false;
     const includeScores = adv.includeScores !== false;
@@ -1722,6 +1773,9 @@ const buildContextMessages = ({ payloadFormat, query, userId, profileFacts, work
         if (connectedModelSummary && adv.includeConnectedModelSummary !== false) {
             sections.push({ section: 'connected_model_summary', title: 'SLM summary', value: connectedModelSummary });
         }
+        else if (activeSummary && adv.includeActiveSummary !== false) {
+            sections.push({ section: 'active_summary', title: 'Active summary', value: activeSummary });
+        }
         sections.push({
             section: 'working_memory',
             title: 'Working memory',
@@ -1790,6 +1844,9 @@ const buildContextMessages = ({ payloadFormat, query, userId, profileFacts, work
     if (connectedModelSummary && adv.includeConnectedModelSummary !== false) {
         sections.push({ section: 'connected_model_summary', title: 'SLM summary', value: connectedModelSummary });
     }
+    else if (activeSummary && adv.includeActiveSummary !== false) {
+        sections.push({ section: 'active_summary', title: 'Active summary', value: activeSummary });
+    }
     sections.push({
         section: 'working_memory',
         title: 'Working memory',
@@ -2113,7 +2170,14 @@ class Mem0Memory {
                             description: 'Define se o SLM resume working memory/tool state, memórias vetoriais, ambos ou nada.',
                         },
                         { displayName: 'Máximo de Caracteres de Entrada do SLM', name: 'connectedModelSummaryInputMaxChars', type: 'number', default: 3000 },
-                        { displayName: 'Máximo de Caracteres do Resumo do SLM', name: 'connectedModelSummaryMaxChars', type: 'number', default: 900 },
+                        { displayName: 'Máximo de Caracteres do Resumo do SLM', name: 'connectedModelSummaryMaxChars', type: 'number', default: 1200 },
+                        { displayName: 'Incluir Active Summary', name: 'includeActiveSummary', type: 'boolean', default: true, description: 'Carrega o resumo ativo persistente da thread/sessão quando disponível.' },
+                        { displayName: 'Persistir Active Summary', name: 'persistActiveSummary', type: 'boolean', default: true, description: 'Salva o resumo atualizado pelo SLM para continuar a conversa depois de minutos, horas ou dias.' },
+                        { displayName: 'Máximo de Caracteres do Active Summary', name: 'activeSummaryMaxChars', type: 'number', default: 1600 },
+                        { displayName: 'Retenção do Active Summary (Dias)', name: 'activeSummaryRetentionDays', type: 'number', default: 30, description: 'Use 0 para não expirar pelo node. Não é cache técnico; é memória ativa persistente.' },
+                        { displayName: 'Ativar Cache Técnico do Resumo SLM', name: 'enableTransientSummaryCache', type: 'boolean', default: true, description: 'Evita chamadas duplicadas ao SLM para o mesmo pacote de contexto. Não substitui memória persistente.' },
+                        { displayName: 'TTL do Cache Técnico SLM (Segundos)', name: 'transientSummaryCacheTTLSeconds', type: 'number', default: 300 },
+                        { displayName: 'Máximo de Itens no Cache Técnico SLM', name: 'transientSummaryCacheMaxItems', type: 'number', default: 50 },
                         { displayName: 'Máximo de Fatos no Resumo', name: 'summaryMaxFacts', type: 'number', default: 4 },
                         { displayName: 'Incluir Scores', name: 'includeScores', type: 'boolean', default: true },
                         { displayName: 'Incluir Diagnóstico', name: 'includeDiagnostics', type: 'boolean', default: false },
@@ -2768,11 +2832,20 @@ class Mem0Memory {
             vectorMemories,
             maxItems: adv.compressionMaxItems || 6,
         });
+        const loadedActiveSummary = readActiveSummary(store, key, adv);
         let connectedModelSummary = '';
+        const summaryDiagnostics = {
+            source: 'none',
+            transientCacheHit: false,
+            activeSummaryLoaded: Boolean(loadedActiveSummary),
+            activeSummaryUpdated: false,
+            summaryChars: 0,
+        };
         if (connectedLanguageModel && typeof connectedLanguageModel.invoke === 'function' && adv.includeSummary !== false && adv.includeConnectedModelSummary !== false) {
             try {
                 const summaryInput = buildConnectedModelSummaryInput({
                     query,
+                    activeSummary: loadedActiveSummary,
                     profileFacts,
                     workingMemory,
                     decisionState,
@@ -2784,13 +2857,47 @@ class Mem0Memory {
                     highlights,
                     adv,
                 });
-                connectedModelSummary = await invokeConnectedModelSummary(connectedLanguageModel, summaryInput, adv);
+                const cacheKey = summaryInput ? stableHash({
+                    key,
+                    source: adv.summarySource || 'auto',
+                    input: summaryInput,
+                    max: adv.connectedModelSummaryMaxChars || 1200,
+                }) : '';
+                const cached = cacheKey ? store.connectedModelSummaryCache[cacheKey] : null;
+                const transientCacheEnabled = adv.enableTransientSummaryCache !== false;
+                const transientTtlSeconds = Number(adv.transientSummaryCacheTTLSeconds || adv.connectedModelSummaryCacheTTLSeconds || 300);
+                if (transientCacheEnabled && cached && cached.summary && Date.now() - Number(cached.at || 0) < transientTtlSeconds * 1000) {
+                    connectedModelSummary = cached.summary;
+                    connectedAi.languageModelSummaryCached = true;
+                    summaryDiagnostics.source = 'transient_cache';
+                    summaryDiagnostics.transientCacheHit = true;
+                }
+                else {
+                    connectedModelSummary = await invokeConnectedModelSummary(connectedLanguageModel, summaryInput, adv);
+                    if (cacheKey && connectedModelSummary) {
+                        store.connectedModelSummaryCache[cacheKey] = { summary: connectedModelSummary, at: Date.now() };
+                        const keys = Object.keys(store.connectedModelSummaryCache);
+                        for (const oldKey of keys.slice(0, Math.max(0, keys.length - Number(adv.transientSummaryCacheMaxItems || adv.connectedModelSummaryCacheMaxItems || 50))))
+                            delete store.connectedModelSummaryCache[oldKey];
+                    }
+                    summaryDiagnostics.source = connectedModelSummary ? 'fresh_slm' : 'none';
+                }
                 connectedAi.languageModelSummary = Boolean(connectedModelSummary);
             }
             catch (error) {
                 connectedAi.errors.push(`languageModel.invoke: ${error.message || String(error)}`);
             }
         }
+        if (!connectedModelSummary && loadedActiveSummary) {
+            connectedModelSummary = loadedActiveSummary;
+            summaryDiagnostics.source = 'active_summary';
+        }
+        if (connectedModelSummary) {
+            summaryDiagnostics.summaryChars = String(connectedModelSummary).length;
+            if (summaryDiagnostics.source === 'fresh_slm') {
+                summaryDiagnostics.activeSummaryUpdated = writeActiveSummary(store, key, connectedModelSummary, adv);
+            }
+        }
         const diagnostics = {
             vectorMemories: vectorMemories.length,
             recentMessages: recentMessages.length,
@@ -2799,6 +2906,7 @@ class Mem0Memory {
             project: project || null,
             memoryNamespace: key,
             connectedAi,
+            activeSummary: summaryDiagnostics,
         };
         const contextHealth = deriveContextHealth({
             userId: key,
@@ -2837,6 +2945,7 @@ class Mem0Memory {
             highlights,
             graph,
             diagnostics,
+            activeSummary: loadedActiveSummary,
             connectedModelSummary,
             adv,
         });
@@ -2883,6 +2992,9 @@ class Mem0Memory {
                 includeOperationalState: adv.includeOperationalState !== false,
                 includeActionLedger: adv.includeActionLedger !== false,
                 includeEntityTimeline: adv.includeEntityTimeline !== false,
+                includeActiveSummary: adv.includeActiveSummary !== false,
+                persistActiveSummary: adv.persistActiveSummary !== false,
+                enableTransientSummaryCache: adv.enableTransientSummaryCache !== false,
             compactForAgent: Boolean(adv.compactForAgent),
             includeWorkingMemory: adv.includeWorkingMemory !== false,
             includeDecisionState: adv.includeDecisionState !== false,
@@ -2891,6 +3003,7 @@ class Mem0Memory {
             },
             context: contextText,
             summary,
+            activeSummary: loadedActiveSummary,
             connectedModelSummary,
             contextHealth,
             contextQualityScore: contextHealth.quality_score,
@@ -2938,6 +3051,7 @@ class Mem0Memory {
                 temboryContext: audit,
                 temboryContextText: contextText,
                 temborySummary: summary,
+                temboryActiveSummary: loadedActiveSummary,
                 temboryConnectedModelSummary: connectedModelSummary,
                 temboryContextHealth: contextHealth,
                 temboryContextQualityScore: contextHealth.quality_score,
@@ -3073,6 +3187,9 @@ exports.__private = {
     compactToolResult,
     compactToolHistoryForAgent,
     compactOperationalStateForAgent,
+    activeSummaryIsFresh,
+    readActiveSummary,
+    writeActiveSummary,
     buildConnectedModelSummaryInput,
     cleanModelSummaryText,
     invokeConnectedModelSummary,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "n8n-nodes-tembory",
-  "version": "1.0.12",
+  "version": "1.0.14",
   "description": "Tembory node for n8n AI Agents with profile, tools, timeline, graph and semantic memory",
   "license": "MIT",
   "homepage": "https://tembory.com",
@@ -48,7 +48,6 @@
     "n8n",
     "tembory",
     "tembory-memory",
-    "elephant-brain",
     "memory",
     "ai",
     "llm",