swarm-engine 1.3.0 → 1.38.0

package/dist/runtime/graph-meta.js

@@ -0,0 +1,437 @@
+// ─── Stop Words ──────────────────────────────────────────────────
+const STOP_WORDS = new Set([
+    'the', 'a', 'an', 'to', 'in', 'for', 'of', 'with', 'and', 'or', 'is', 'are', 'was', 'were',
+    'be', 'been', 'it', 'its', 'this', 'that', 'by', 'from', 'on', 'at', 'as', 'but', 'not', 'no',
+    'do', 'does', 'did', 'have', 'has', 'had', 'will', 'would', 'could', 'should', 'can', 'may',
+    'might', 'shall', 'must', 'need',
+]);
+// ─── MetaPatternSelector ─────────────────────────────────────────
+/**
+ * MAML-inspired few-shot pattern selection using TF-IDF + logistic regression.
+ * Learns from past orchestration outcomes to recommend the best pattern for new tasks.
+ * Persists state to the execution knowledge graph for cross-run learning.
+ */
+export class MetaPatternSelector {
+    graph;
+    tfidfModel = null;
+    patternModels = new Map();
+    examples = [];
+    constructor(graph) {
+        this.graph = graph;
+        this.loadFromGraph();
+    }
+    // ─── Text Processing ──────────────────────────────────────────
+    /**
+     * Tokenize text: lowercase, replace non-word chars with spaces,
+     * split on whitespace, filter tokens ≤ 2 chars, remove stop words.
+     */
+    static tokenize(text) {
+        return text
+            .toLowerCase()
+            .replace(/[^\w]/g, ' ')
+            .split(/\s+/)
+            .filter((t) => t.length > 2 && !STOP_WORDS.has(t));
+    }
+    /**
+     * Build TF-IDF model from a document corpus.
+     * IDF uses smoothed formula: log((N+1)/(df+1)) + 1
+     * Vocabulary capped at maxVocab terms, sorted by document frequency.
+     */
+    static buildTfIdfModel(documents, maxVocab = 500) {
+        const N = documents.length;
+        const tokenizedDocs = documents.map((d) => MetaPatternSelector.tokenize(d));
+        // Count document frequency for each term
+        const dfMap = new Map();
+        for (const tokens of tokenizedDocs) {
+            const unique = new Set(tokens);
+            for (const token of unique) {
+                dfMap.set(token, (dfMap.get(token) ?? 0) + 1);
+            }
+        }
+        // Sort by document frequency (descending), cap at maxVocab
+        const sorted = [...dfMap.entries()]
+            .sort((a, b) => b[1] - a[1])
+            .slice(0, maxVocab);
+        const vocabulary = new Map();
+        const idf = [];
+        for (let i = 0; i < sorted.length; i++) {
+            const [word, df] = sorted[i];
+            vocabulary.set(word, i);
+            idf.push(Math.log((N + 1) / (df + 1)) + 1);
+        }
+        return { vocabulary, idf, documentCount: N };
+    }
+    /**
+     * Vectorize a single text against the TF-IDF model.
+     * Returns L2-normalized vector.
+     */
+    static vectorize(model, text) {
+        const tokens = MetaPatternSelector.tokenize(text);
+        const vec = new Array(model.vocabulary.size).fill(0);
+        // Term frequency
+        const tf = new Map();
+        for (const token of tokens) {
+            tf.set(token, (tf.get(token) ?? 0) + 1);
+        }
+        // TF-IDF
+        for (const [word, count] of tf) {
+            const idx = model.vocabulary.get(word);
+            if (idx !== undefined) {
+                vec[idx] = count * model.idf[idx];
+            }
+        }
+        // L2 normalize
+        const magnitude = Math.sqrt(vec.reduce((sum, v) => sum + v * v, 0));
+        if (magnitude > 0) {
+            for (let i = 0; i < vec.length; i++) {
+                vec[i] /= magnitude;
+            }
+        }
+        return vec;
+    }
+    /**
+     * Cosine similarity between two vectors.
+     * Returns 0 if either vector has zero magnitude.
+     */
+    static cosineSimilarity(a, b) {
+        const len = Math.min(a.length, b.length);
+        let dot = 0;
+        let magA = 0;
+        let magB = 0;
+        for (let i = 0; i < len; i++) {
+            dot += a[i] * b[i];
+            magA += a[i] * a[i];
+            magB += b[i] * b[i];
+        }
+        const denom = Math.sqrt(magA) * Math.sqrt(magB);
+        return denom === 0 ? 0 : dot / denom;
+    }
+    // ─── Feature Engineering ──────────────────────────────────────
+    /**
+     * Build full feature vector: [tfidf..., log(fileCount+1), log(totalLines+1), log(numTestFiles+1), hasReviewPhase ? 1 : 0]
+     */
+    buildFeatureVector(taskDescription, context) {
+        // Rebuild TF-IDF model if needed
+        if (!this.tfidfModel) {
+            const docs = this.examples.length > 0
+                ? this.getStoredTaskDescriptions()
+                : [taskDescription];
+            // Only build if we actually have documents with tokens
+            if (docs.length > 0) {
+                this.tfidfModel = MetaPatternSelector.buildTfIdfModel(docs);
+            }
+        }
+        const tfidfVec = this.tfidfModel
+            ? MetaPatternSelector.vectorize(this.tfidfModel, taskDescription)
+            : [];
+        // Contextual features
+        const fileCount = context?.fileCount ?? 0;
+        const totalLines = context?.totalLines ?? 0;
+        const numTestFiles = context?.numTestFiles ?? 0;
+        const hasReview = context?.hasReviewPhase ? 1 : 0;
+        return [
+            ...tfidfVec,
+            Math.log(fileCount + 1),
+            Math.log(totalLines + 1),
+            Math.log(numTestFiles + 1),
+            hasReview,
+        ];
+    }
+    // ─── Logistic Regression ──────────────────────────────────────
+    /**
+     * Predict success probability: σ(w·x + b)
+     * Sigmoid input clamped to [-500, 500] for numerical stability.
+     */
+    predict(model, features) {
+        let z = model.bias;
+        const len = Math.min(model.weights.length, features.length);
+        for (let i = 0; i < len; i++) {
+            z += model.weights[i] * features[i];
+        }
+        // Clamp to avoid overflow
+        z = Math.max(-500, Math.min(500, z));
+        return 1 / (1 + Math.exp(-z));
+    }
+    /**
+     * Update model weights with one example via online SGD.
+     * error = predict(features) - label
+     * w[i] -= lr * (error * features[i] + lambda * w[i])
+     * bias -= lr * error
+     */
+    updateModel(model, features, label) {
+        const prediction = this.predict(model, features);
+        const error = prediction - label;
+        const len = Math.min(model.weights.length, features.length);
+        for (let i = 0; i < len; i++) {
+            model.weights[i] -= model.learnRate * (error * features[i] + model.lambda * model.weights[i]);
+        }
+        model.bias -= model.learnRate * error;
+        model.totalExamples++;
+        if (label === 1)
+            model.positiveExamples++;
+    }
+    /**
+     * MAML inner loop: adapt model to new task using K most similar historical examples.
+     * 1. Find K most similar examples by cosine similarity on feature vectors
+     * 2. Clone the base model (deep copy weights)
+     * 3. Run K gradient steps on those examples with inner LR=0.05
+     * 4. Return adapted model
+     */
+    adaptToNewTask(baseModel, newTaskFeatures, K = 3) {
+        // Find K most similar examples for this pattern
+        const patternExamples = this.examples.filter((e) => e.pattern === baseModel.pattern);
+        const scored = patternExamples.map((ex) => ({
+            example: ex,
+            similarity: MetaPatternSelector.cosineSimilarity(newTaskFeatures, ex.features),
+        }));
+        scored.sort((a, b) => b.similarity - a.similarity);
+        const topK = scored.slice(0, K);
+        // Deep clone the base model
+        const adapted = {
+            pattern: baseModel.pattern,
+            weights: [...baseModel.weights],
+            bias: baseModel.bias,
+            learnRate: 0.05, // Inner loop learning rate
+            lambda: baseModel.lambda,
+            totalExamples: baseModel.totalExamples,
+            positiveExamples: baseModel.positiveExamples,
+        };
+        // Run gradient steps on the K most similar examples
+        for (const { example } of topK) {
+            this.updateModel(adapted, example.features, example.label);
+        }
+        return adapted;
+    }
+    // ─── Pattern Selection ────────────────────────────────────────
+    /**
+     * Main entry point: select the best pattern for a task.
+     * For each known pattern, predicts success probability using adapted models.
+     * Returns the highest-confidence pattern with alternatives.
+     */
+    selectPattern(taskDescription, context) {
+        const features = this.buildFeatureVector(taskDescription, context);
+        // If we have no data, rebuild TF-IDF from stored descriptions
+        if (!this.tfidfModel && this.examples.length > 0) {
+            const docs = this.getStoredTaskDescriptions();
+            if (docs.length > 0) {
+                this.tfidfModel = MetaPatternSelector.buildTfIdfModel(docs);
+            }
+        }
+        // Collect known patterns from graph and from examples
+        const knownPatterns = this.getKnownPatterns();
+        if (knownPatterns.length === 0) {
+            return {
+                pattern: 'hybrid',
+                confidence: 0.5,
+                reasoning: 'No historical data available; defaulting to hybrid pattern',
+                alternatives: [],
+            };
+        }
+        // Score each pattern
+        const scores = [];
+        for (const pattern of knownPatterns) {
+            let model = this.patternModels.get(pattern);
+            if (!model) {
+                model = this.createPatternModel(pattern, features.length);
+                this.patternModels.set(pattern, model);
+            }
+            // Ensure weights match feature vector length
+            this.ensureWeightSize(model, features.length);
+            let prediction;
+            const patternExampleCount = this.examples.filter((e) => e.pattern === pattern).length;
+            if (patternExampleCount >= 3) {
+                const adapted = this.adaptToNewTask(model, features);
+                prediction = this.predict(adapted, features);
+            }
+            else {
+                prediction = this.predict(model, features);
+            }
+            scores.push({ pattern, confidence: prediction });
+        }
+        // Sort by confidence descending
+        scores.sort((a, b) => b.confidence - a.confidence);
+        const best = scores[0];
+        const alternatives = scores.slice(1).map((s) => ({ pattern: s.pattern, confidence: s.confidence }));
+        const exampleCount = this.examples.filter((e) => e.pattern === best.pattern).length;
+        const reasoning = exampleCount >= 3
+            ? `MAML-adapted prediction from ${exampleCount} historical examples`
+            : exampleCount > 0
+                ? `Base model prediction from ${exampleCount} examples (< 3, no adaptation)`
+                : 'Cold start — no examples for this pattern yet';
+        return {
+            pattern: best.pattern,
+            confidence: best.confidence,
+            reasoning,
+            alternatives,
+        };
+    }
+    // ─── Online Learning ──────────────────────────────────────────
+    /**
+     * Record orchestration outcome for online learning.
+     * Fetches task description and pattern from the graph, builds features,
+     * updates the pattern model, and stores the example.
+     */
+    updateFromOutcome(orchestrationId, success) {
+        const orchNode = this.graph.getNode(orchestrationId);
+        if (!orchNode)
+            return;
+        const taskDescription = orchNode.properties.taskDescription || orchNode.label;
+        const pattern = orchNode.properties.pattern;
+        if (!pattern)
+            return;
+        const features = this.buildFeatureVector(taskDescription, {
+            fileCount: orchNode.properties.fileCount,
+            totalLines: orchNode.properties.totalLines,
+            numTestFiles: orchNode.properties.numTestFiles,
+            hasReviewPhase: orchNode.properties.hasReviewPhase,
+        });
+        const label = success ? 1 : 0;
+        // Store example
+        const example = { features, label, pattern };
+        this.examples.push(example);
+        // Update the pattern model
+        let model = this.patternModels.get(pattern);
+        if (!model) {
+            model = this.createPatternModel(pattern, features.length);
+            this.patternModels.set(pattern, model);
+        }
+        this.ensureWeightSize(model, features.length);
+        this.updateModel(model, features, label);
+        // Rebuild TF-IDF with the new document
+        const docs = this.getStoredTaskDescriptions();
+        docs.push(taskDescription);
+        this.tfidfModel = MetaPatternSelector.buildTfIdfModel(docs);
+        // Persist
+        this.saveToGraph();
+    }
+    // ─── Persistence ──────────────────────────────────────────────
+    saveToGraph() {
+        // Store TF-IDF model as a heuristic node
+        if (this.tfidfModel) {
+            const vocabObj = {};
+            for (const [word, idx] of this.tfidfModel.vocabulary) {
+                vocabObj[word] = idx;
+            }
+            this.graph.addNode({
+                id: 'meta:tfidf',
+                type: 'heuristic',
+                label: 'MetaPatternSelector TF-IDF Model',
+                properties: {
+                    vocabulary: vocabObj,
+                    idf: this.tfidfModel.idf,
+                    documentCount: this.tfidfModel.documentCount,
+                },
+            });
+        }
+        // Store pattern models
+        for (const [pattern, model] of this.patternModels) {
+            this.graph.addNode({
+                id: `meta:pattern:${pattern}`,
+                type: 'heuristic',
+                label: `MetaPatternSelector model: ${pattern}`,
+                properties: {
+                    pattern: model.pattern,
+                    weights: model.weights,
+                    bias: model.bias,
+                    learnRate: model.learnRate,
+                    lambda: model.lambda,
+                    totalExamples: model.totalExamples,
+                    positiveExamples: model.positiveExamples,
+                },
+            });
+        }
+        // Store examples (capped to last 1000 to avoid bloat)
+        const capped = this.examples.slice(-1000);
+        this.graph.addNode({
+            id: 'meta:examples',
+            type: 'heuristic',
+            label: 'MetaPatternSelector examples',
+            properties: {
+                examples: capped,
+            },
+        });
+    }
+    loadFromGraph() {
+        // Load TF-IDF model
+        const tfidfNode = this.graph.getNode('meta:tfidf');
+        if (tfidfNode) {
+            const vocabObj = tfidfNode.properties.vocabulary;
+            const idf = tfidfNode.properties.idf;
+            const documentCount = tfidfNode.properties.documentCount;
+            if (vocabObj && idf && documentCount !== undefined) {
+                const vocabulary = new Map();
+                for (const [word, idx] of Object.entries(vocabObj)) {
+                    vocabulary.set(word, idx);
+                }
+                this.tfidfModel = { vocabulary, idf, documentCount };
+            }
+        }
+        // Load pattern models
+        const heuristicNodes = this.graph.getNodesByType('heuristic');
+        for (const node of heuristicNodes) {
+            if (node.id.startsWith('meta:pattern:')) {
+                const props = node.properties;
+                if (props.pattern && props.weights) {
+                    this.patternModels.set(props.pattern, {
+                        pattern: props.pattern,
+                        weights: props.weights,
+                        bias: props.bias ?? 0,
+                        learnRate: props.learnRate ?? 0.01,
+                        lambda: props.lambda ?? 0.001,
+                        totalExamples: props.totalExamples ?? 0,
+                        positiveExamples: props.positiveExamples ?? 0,
+                    });
+                }
+            }
+        }
+        // Load examples
+        const examplesNode = this.graph.getNode('meta:examples');
+        if (examplesNode) {
+            const stored = examplesNode.properties.examples;
+            if (Array.isArray(stored)) {
+                this.examples = stored;
+            }
+        }
+    }
+    // ─── Helpers ──────────────────────────────────────────────────
+    createPatternModel(pattern, featureSize) {
+        return {
+            pattern,
+            weights: new Array(featureSize).fill(0),
+            bias: 0,
+            learnRate: 0.01,
+            lambda: 0.001,
+            totalExamples: 0,
+            positiveExamples: 0,
+        };
+    }
+    ensureWeightSize(model, size) {
+        while (model.weights.length < size) {
+            model.weights.push(0);
+        }
+    }
+    getKnownPatterns() {
+        const patterns = new Set();
+        // From pattern models
+        for (const pattern of this.patternModels.keys()) {
+            patterns.add(pattern);
+        }
+        // From examples
+        for (const ex of this.examples) {
+            patterns.add(ex.pattern);
+        }
+        // From graph pattern nodes
+        const patternNodes = this.graph.getNodesByType('pattern');
+        for (const node of patternNodes) {
+            patterns.add(node.label);
+        }
+        return [...patterns];
+    }
+    getStoredTaskDescriptions() {
+        const orchNodes = this.graph.getNodesByType('orchestration');
+        return orchNodes
+            .map((n) => n.properties.taskDescription || n.label)
+            .filter((desc) => desc && desc.length > 0);
+    }
+}
+//# sourceMappingURL=graph-meta.js.map

package/dist/runtime/graph-meta.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"graph-meta.js","sourceRoot":"","sources":["../../src/runtime/graph-meta.ts"],"names":[],"mappings":"AAEA,oEAAoE;AAEpE,MAAM,UAAU,GAAG,IAAI,GAAG,CAAC;IACzB,KAAK,EAAE,GAAG,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,IAAI,EAAE,IAAI,EAAE,KAAK,EAAE,KAAK,EAAE,MAAM;IAC1F,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,KAAK,EAAE,MAAM,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,KAAK,EAAE,KAAK,EAAE,IAAI;IAC7F,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,KAAK,EAAE,MAAM,EAAE,OAAO,EAAE,OAAO,EAAE,QAAQ,EAAE,KAAK,EAAE,KAAK;IAC3F,OAAO,EAAE,OAAO,EAAE,MAAM,EAAE,MAAM;CACjC,CAAC,CAAC;AAiCH,oEAAoE;AAEpE;;;;GAIG;AACH,MAAM,OAAO,mBAAmB;IAKV;IAJZ,UAAU,GAAsB,IAAI,CAAC;IACrC,aAAa,GAA8B,IAAI,GAAG,EAAE,CAAC;IACrD,QAAQ,GAAkB,EAAE,CAAC;IAErC,YAAoB,KAAqB;QAArB,UAAK,GAAL,KAAK,CAAgB;QACvC,IAAI,CAAC,aAAa,EAAE,CAAC;IACvB,CAAC;IAED,iEAAiE;IAEjE;;;OAGG;IACH,MAAM,CAAC,QAAQ,CAAC,IAAY;QAC1B,OAAO,IAAI;aACR,WAAW,EAAE;aACb,OAAO,CAAC,QAAQ,EAAE,GAAG,CAAC;aACtB,KAAK,CAAC,KAAK,CAAC;aACZ,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,GAAG,CAAC,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;IACvD,CAAC;IAED;;;;OAIG;IACH,MAAM,CAAC,eAAe,CAAC,SAAmB,EAAE,QAAQ,GAAG,GAAG;QACxD,MAAM,CAAC,GAAG,SAAS,CAAC,MAAM,CAAC;QAC3B,MAAM,aAAa,GAAG,SAAS,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,mBAAmB,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC;QAE5E,yCAAyC;QACzC,MAAM,KAAK,GAAG,IAAI,GAAG,EAAkB,CAAC;QACxC,KAAK,MAAM,MAAM,IAAI,aAAa,EAAE,CAAC;YACnC,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,MAAM,CAAC,CAAC;YAC/B,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;gBAC3B,KAAK,CAAC,GAAG,CAAC,KAAK,EAAE,CAAC,KAAK,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;YAChD,CAAC;QACH,CAAC;QAED,2DAA2D;QAC3D,MAAM,MAAM,GAAG,CAAC,GAAG,KAAK,CAAC,OAAO,EAAE,CAAC;aAChC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;aAC3B,KAAK,CAAC,CAAC,EAAE,QAAQ,CAAC,CAAC;QAEtB,MAAM,UAAU,GAAG,IAAI,GAAG,EAAkB,CAAC;QAC7C,MAAM,GAAG,GAAa,EAAE,CAAC;QACzB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;YACvC,MAAM,CAAC,IAAI,EAAE,EAAE,CAAC,GAAG,MAAM,CAAC,CAAC,CAAC,CAAC;YAC7B,UAAU,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC;YACxB,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;QAC7C,CAAC;QAED,OAAO,EAAE,UAAU,EAAE,GAAG,EAAE,aAAa,EAAE,CAAC,EAAE,CAAC;IAC/C,CAAC;IAED;;;OAGG;IACH,MAAM,CAAC,SAAS,CAAC,KAAiB,EAAE,IAAY;QAC9C,MAAM,MAAM,GAAG,mBAAmB,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC;QAClD,MAAM,GAAG,GAAG,IAAI,KAAK,CAAC,KAAK,CAAC,UAAU,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;QAErD,iBAAiB;QACjB,MAAM,EAAE,GAAG,IAAI,GAAG,EAAkB,CAAC;QACrC,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;YAC3B,EAAE,CAAC,GAAG,CAAC,KAAK,EAAE,CAAC,EAAE,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;QAC1C,CAAC;QAED,SAAS;QACT,KAAK,MAAM,CAAC,IAAI,EAAE,KAAK,CAAC,IAAI,EAAE,EAAE,CAAC;YAC/B,MAAM,GAAG,GAAG,KAAK,CAAC,UAAU,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;YACvC,IAAI,GAAG,KAAK,SAAS,EAAE,CAAC;gBACtB,GAAG,CAAC,GAAG,CAAC,GAAG,KAAK,GAAG,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;YACpC,CAAC;QACH,CAAC;QAED,eAAe;QACf,MAAM,SAAS,GAAG,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,GAAG,GAAG,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;QACpE,IAAI,SAAS,GAAG,CAAC,EAAE,CAAC;YAClB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,GAAG,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;gBACpC,GAAG,CAAC,CAAC,CAAC,IAAI,SAAS,CAAC;YACtB,CAAC;QACH,CAAC;QAED,OAAO,GAAG,CAAC;IACb,CAAC;IAED;;;OAGG;IACH,MAAM,CAAC,gBAAgB,CAAC,CAAW,EAAE,CAAW;QAC9C,MAAM,GAAG,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,MAAM,EAAE,CAAC,CAAC,MAAM,CAAC,CAAC;QACzC,IAAI,GAAG,GAAG,CAAC,CAAC;QACZ,IAAI,IAAI,GAAG,CAAC,CAAC;QACb,IAAI,IAAI,GAAG,CAAC,CAAC;QACb,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC;YAC7B,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;YACnB,IAAI,IAAI,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;YACpB,IAAI,IAAI,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;QACtB,CAAC;QACD,MAAM,KAAK,GAAG,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAChD,OAAO,KAAK,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,GAAG,KAAK,CAAC;IACvC,CAAC;IAED,iEAAiE;IAEjE;;OAEG;IACH,kBAAkB,CAChB,eAAuB,EACvB,OAAsG;QAEtG,iCAAiC;QACjC,IAAI,CAAC,IAAI,CAAC,UAAU,EAAE,CAAC;YACrB,MAAM,IAAI,GAAG,IAAI,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC;gBACnC,CAAC,CAAC,IAAI,CAAC,yBAAyB,EAAE;gBAClC,CAAC,CAAC,CAAC,eAAe,CAAC,CAAC;YACtB,uDAAuD;YACvD,IAAI,IAAI,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBACpB,IAAI,CAAC,UAAU,GAAG,mBAAmB,CAAC,eAAe,CAAC,IAAI,CAAC,CAAC;YAC9D,CAAC;QACH,CAAC;QAED,MAAM,QAAQ,GAAG,IAAI,CAAC,UAAU;YAC9B,CAAC,CAAC,mBAAmB,CAAC,SAAS,CAAC,IAAI,CAAC,UAAU,EAAE,eAAe,CAAC;YACjE,CAAC,CAAC,EAAE,CAAC;QAEP,sBAAsB;QACtB,MAAM,SAAS,GAAG,OAAO,EAAE,SAAS,IAAI,CAAC,CAAC;QAC1C,MAAM,UAAU,GAAG,OAAO,EAAE,UAAU,IAAI,CAAC,CAAC;QAC5C,MAAM,YAAY,GAAG,OAAO,EAAE,YAAY,IAAI,CAAC,CAAC;QAChD,MAAM,SAAS,GAAG,OAAO,EAAE,cAAc,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QAElD,OAAO;YACL,GAAG,QAAQ;YACX,IAAI,CAAC,GAAG,CAAC,SAAS,GAAG,CAAC,CAAC;YACvB,IAAI,CAAC,GAAG,CAAC,UAAU,GAAG,CAAC,CAAC;YACxB,IAAI,CAAC,GAAG,CAAC,YAAY,GAAG,CAAC,CAAC;YAC1B,SAAS;SACV,CAAC;IACJ,CAAC;IAED,iEAAiE;IAEjE;;;OAGG;IACH,OAAO,CAAC,KAAmB,EAAE,QAAkB;QAC7C,IAAI,CAAC,GAAG,KAAK,CAAC,IAAI,CAAC;QACnB,MAAM,GAAG,GAAG,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,OAAO,CAAC,MAAM,EAAE,QAAQ,CAAC,MAAM,CAAC,CAAC;QAC5D,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC;YAC7B,CAAC,IAAI,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAC;QACtC,CAAC;QACD,0BAA0B;QAC1B,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,IAAI,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,CAAC;QACrC,OAAO,CAAC,GAAG,CAAC,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IAChC,CAAC;IAED;;;;;OAKG;IACH,WAAW,CAAC,KAAmB,EAAE,QAAkB,EAAE,KAAY;QAC/D,MAAM,UAAU,GAAG,IAAI,CAAC,OAAO,CAAC,KAAK,EAAE,QAAQ,CAAC,CAAC;QACjD,MAAM,KAAK,GAAG,UAAU,GAAG,KAAK,CAAC;QAEjC,MAAM,GAAG,GAAG,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,OAAO,CAAC,MAAM,EAAE,QAAQ,CAAC,MAAM,CAAC,CAAC;QAC5D,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC;YAC7B,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,SAAS,GAAG,CAAC,KAAK,GAAG,QAAQ,CAAC,CAAC,CAAC,GAAG,KAAK,CAAC,MAAM,GAAG,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,CAAC;QAChG,CAAC;QACD,KAAK,CAAC,IAAI,IAAI,KAAK,CAAC,SAAS,GAAG,KAAK,CAAC;QAEtC,KAAK,CAAC,aAAa,EAAE,CAAC;QACtB,IAAI,KAAK,KAAK,CAAC;YAAE,KAAK,CAAC,gBAAgB,EAAE,CAAC;IAC5C,CAAC;IAED;;;;;;OAMG;IACH,cAAc,CAAC,SAAuB,EAAE,eAAyB,EAAE,CAAC,GAAG,CAAC;QACtE,gDAAgD;QAChD,MAAM,eAAe,GAAG,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,OAAO,KAAK,SAAS,CAAC,OAAO,CAAC,CAAC;QAErF,MAAM,MAAM,GAAG,eAAe,CAAC,GAAG,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,CAAC;YAC1C,OAAO,EAAE,EAAE;YACX,UAAU,EAAE,mBAAmB,CAAC,gBAAgB,CAAC,eAAe,EAAE,EAAE,CAAC,QAAQ,CAAC;SAC/E,CAAC,CAAC,CAAC;QACJ,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,UAAU,GAAG,CAAC,CAAC,UAAU,CAAC,CAAC;QACnD,MAAM,IAAI,GAAG,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;QAEhC,4BAA4B;QAC5B,MAAM,OAAO,GAAiB;YAC5B,OAAO,EAAE,SAAS,CAAC,OAAO;YAC1B,OAAO,EAAE,CAAC,GAAG,SAAS,CAAC,OAAO,CAAC;YAC/B,IAAI,EAAE,SAAS,CAAC,IAAI;YACpB,SAAS,EAAE,IAAI,EAAE,2BAA2B;YAC5C,MAAM,EAAE,SAAS,CAAC,MAAM;YACxB,aAAa,EAAE,SAAS,CAAC,aAAa;YACtC,gBAAgB,EAAE,SAAS,CAAC,gBAAgB;SAC7C,CAAC;QAEF,oDAAoD;QACpD,KAAK,MAAM,EAAE,OAAO,EAAE,IAAI,IAAI,EAAE,CAAC;YAC/B,IAAI,CAAC,WAAW,CAAC,OAAO,EAAE,OAAO,CAAC,QAAQ,EAAE,OAAO,CAAC,KAAK,CAAC,CAAC;QAC7D,CAAC;QAED,OAAO,OAAO,CAAC;IACjB,CAAC;IAED,iEAAiE;IAEjE;;;;OAIG;IACH,aAAa,CACX,eAAuB,EACvB,OAAsG;QAEtG,MAAM,QAAQ,GAAG,IAAI,CAAC,kBAAkB,CAAC,eAAe,EAAE,OAAO,CAAC,CAAC;QAEnE,8DAA8D;QAC9D,IAAI,CAAC,IAAI,CAAC,UAAU,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACjD,MAAM,IAAI,GAAG,IAAI,CAAC,yBAAyB,EAAE,CAAC;YAC9C,IAAI,IAAI,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBACpB,IAAI,CAAC,UAAU,GAAG,mBAAmB,CAAC,eAAe,CAAC,IAAI,CAAC,CAAC;YAC9D,CAAC;QACH,CAAC;QAED,sDAAsD;QACtD,MAAM,aAAa,GAAG,IAAI,CAAC,gBAAgB,EAAE,CAAC;QAE9C,IAAI,aAAa,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YAC/B,OAAO;gBACL,OAAO,EAAE,QAAQ;gBACjB,UAAU,EAAE,GAAG;gBACf,SAAS,EAAE,4DAA4D;gBACvE,YAAY,EAAE,EAAE;aACjB,CAAC;QACJ,CAAC;QAED,qBAAqB;QACrB,MAAM,MAAM,GAAmD,EAAE,CAAC;QAElE,KAAK,MAAM,OAAO,IAAI,aAAa,EAAE,CAAC;YACpC,IAAI,KAAK,GAAG,IAAI,CAAC,aAAa,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;YAC5C,IAAI,CAAC,KAAK,EAAE,CAAC;gBACX,KAAK,GAAG,IAAI,CAAC,kBAAkB,CAAC,OAAO,EAAE,QAAQ,CAAC,MAAM,CAAC,CAAC;gBAC1D,IAAI,CAAC,aAAa,CAAC,GAAG,CAAC,OAAO,EAAE,KAAK,CAAC,CAAC;YACzC,CAAC;YAED,6CAA6C;YAC7C,IAAI,CAAC,gBAAgB,CAAC,KAAK,EAAE,QAAQ,CAAC,MAAM,CAAC,CAAC;YAE9C,IAAI,UAAkB,CAAC;YACvB,MAAM,mBAAmB,GAAG,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,OAAO,KAAK,OAAO,CAAC,CAAC,MAAM,CAAC;YAEtF,IAAI,mBAAmB,IAAI,CAAC,EAAE,CAAC;gBAC7B,MAAM,OAAO,GAAG,IAAI,CAAC,cAAc,CAAC,KAAK,EAAE,QAAQ,CAAC,CAAC;gBACrD,UAAU,GAAG,IAAI,CAAC,OAAO,CAAC,OAAO,EAAE,QAAQ,CAAC,CAAC;YAC/C,CAAC;iBAAM,CAAC;gBACN,UAAU,GAAG,IAAI,CAAC,OAAO,CAAC,KAAK,EAAE,QAAQ,CAAC,CAAC;YAC7C,CAAC;YAED,MAAM,CAAC,IAAI,CAAC,EAAE,OAAO,EAAE,UAAU,EAAE,UAAU,EAAE,CAAC,CAAC;QACnD,CAAC;QAED,gCAAgC;QAChC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,UAAU,GAAG,CAAC,CAAC,UAAU,CAAC,CAAC;QAEnD,MAAM,IAAI,GAAG,MAAM,CAAC,CAAC,CAAC,CAAC;QACvB,MAAM,YAAY,GAAG,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,EAAE,OAAO,EAAE,CAAC,CAAC,OAAO,EAAE,UAAU,EAAE,CAAC,CAAC,UAAU,EAAE,CAAC,CAAC,CAAC;QAEpG,MAAM,YAAY,GAAG,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,OAAO,KAAK,IAAI,CAAC,OAAO,CAAC,CAAC,MAAM,CAAC;QACpF,MAAM,SAAS,GAAG,YAAY,IAAI,CAAC;YACjC,CAAC,CAAC,gCAAgC,YAAY,sBAAsB;YACpE,CAAC,CAAC,YAAY,GAAG,CAAC;gBAChB,CAAC,CAAC,8BAA8B,YAAY,gCAAgC;gBAC5E,CAAC,CAAC,+CAA+C,CAAC;QAEtD,OAAO;YACL,OAAO,EAAE,IAAI,CAAC,OAAO;YACrB,UAAU,EAAE,IAAI,CAAC,UAAU;YAC3B,SAAS;YACT,YAAY;SACb,CAAC;IACJ,CAAC;IAED,iEAAiE;IAEjE;;;;OAIG;IACH,iBAAiB,CAAC,eAAuB,EAAE,OAAgB;QACzD,MAAM,QAAQ,GAAG,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,eAAe,CAAC,CAAC;QACrD,IAAI,CAAC,QAAQ;YAAE,OAAO;QAEtB,MAAM,eAAe,GAAI,QAAQ,CAAC,UAAU,CAAC,eAA0B,IAAI,QAAQ,CAAC,KAAK,CAAC;QAC1F,MAAM,OAAO,GAAG,QAAQ,CAAC,UAAU,CAAC,OAAiB,CAAC;QACtD,IAAI,CAAC,OAAO;YAAE,OAAO;QAErB,MAAM,QAAQ,GAAG,IAAI,CAAC,kBAAkB,CAAC,eAAe,EAAE;YACxD,SAAS,EAAE,QAAQ,CAAC,UAAU,CAAC,SAA+B;YAC9D,UAAU,EAAE,QAAQ,CAAC,UAAU,CAAC,UAAgC;YAChE,YAAY,EAAE,QAAQ,CAAC,UAAU,CAAC,YAAkC;YACpE,cAAc,EAAE,QAAQ,CAAC,UAAU,CAAC,cAAqC;SAC1E,CAAC,CAAC;QAEH,MAAM,KAAK,GAAU,OAAO,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QAErC,gBAAgB;QAChB,MAAM,OAAO,GAAgB,EAAE,QAAQ,EAAE,KAAK,EAAE,OAAO,EAAE,CAAC;QAC1D,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;QAE5B,2BAA2B;QAC3B,IAAI,KAAK,GAAG,IAAI,CAAC,aAAa,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;QAC5C,IAAI,CAAC,KAAK,EAAE,CAAC;YACX,KAAK,GAAG,IAAI,CAAC,kBAAkB,CAAC,OAAO,EAAE,QAAQ,CAAC,MAAM,CAAC,CAAC;YAC1D,IAAI,CAAC,aAAa,CAAC,GAAG,CAAC,OAAO,EAAE,KAAK,CAAC,CAAC;QACzC,CAAC;QACD,IAAI,CAAC,gBAAgB,CAAC,KAAK,EAAE,QAAQ,CAAC,MAAM,CAAC,CAAC;QAC9C,IAAI,CAAC,WAAW,CAAC,KAAK,EAAE,QAAQ,EAAE,KAAK,CAAC,CAAC;QAEzC,uCAAuC;QACvC,MAAM,IAAI,GAAG,IAAI,CAAC,yBAAyB,EAAE,CAAC;QAC9C,IAAI,CAAC,IAAI,CAAC,eAAe,CAAC,CAAC;QAC3B,IAAI,CAAC,UAAU,GAAG,mBAAmB,CAAC,eAAe,CAAC,IAAI,CAAC,CAAC;QAE5D,UAAU;QACV,IAAI,CAAC,WAAW,EAAE,CAAC;IACrB,CAAC;IAED,iEAAiE;IAEzD,WAAW;QACjB,yCAAyC;QACzC,IAAI,IAAI,CAAC,UAAU,EAAE,CAAC;YACpB,MAAM,QAAQ,GAA2B,EAAE,CAAC;YAC5C,KAAK,MAAM,CAAC,IAAI,EAAE,GAAG,CAAC,IAAI,IAAI,CAAC,UAAU,CAAC,UAAU,EAAE,CAAC;gBACrD,QAAQ,CAAC,IAAI,CAAC,GAAG,GAAG,CAAC;YACvB,CAAC;YACD,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC;gBACjB,EAAE,EAAE,YAAY;gBAChB,IAAI,EAAE,WAAW;gBACjB,KAAK,EAAE,kCAAkC;gBACzC,UAAU,EAAE;oBACV,UAAU,EAAE,QAAQ;oBACpB,GAAG,EAAE,IAAI,CAAC,UAAU,CAAC,GAAG;oBACxB,aAAa,EAAE,IAAI,CAAC,UAAU,CAAC,aAAa;iBAC7C;aACF,CAAC,CAAC;QACL,CAAC;QAED,uBAAuB;QACvB,KAAK,MAAM,CAAC,OAAO,EAAE,KAAK,CAAC,IAAI,IAAI,CAAC,aAAa,EAAE,CAAC;YAClD,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC;gBACjB,EAAE,EAAE,gBAAgB,OAAO,EAAE;gBAC7B,IAAI,EAAE,WAAW;gBACjB,KAAK,EAAE,8BAA8B,OAAO,EAAE;gBAC9C,UAAU,EAAE;oBACV,OAAO,EAAE,KAAK,CAAC,OAAO;oBACtB,OAAO,EAAE,KAAK,CAAC,OAAO;oBACtB,IAAI,EAAE,KAAK,CAAC,IAAI;oBAChB,SAAS,EAAE,KAAK,CAAC,SAAS;oBAC1B,MAAM,EAAE,KAAK,CAAC,MAAM;oBACpB,aAAa,EAAE,KAAK,CAAC,aAAa;oBAClC,gBAAgB,EAAE,KAAK,CAAC,gBAAgB;iBACzC;aACF,CAAC,CAAC;QACL,CAAC;QAED,sDAAsD;QACtD,MAAM,MAAM,GAAG,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,CAAC;QAC1C,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC;YACjB,EAAE,EAAE,eAAe;YACnB,IAAI,EAAE,WAAW;YACjB,KAAK,EAAE,8BAA8B;YACrC,UAAU,EAAE;gBACV,QAAQ,EAAE,MAAM;aACjB;SACF,CAAC,CAAC;IACL,CAAC;IAEO,aAAa;QACnB,oBAAoB;QACpB,MAAM,SAAS,GAAG,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,YAAY,CAAC,CAAC;QACnD,IAAI,SAAS,EAAE,CAAC;YACd,MAAM,QAAQ,GAAG,SAAS,CAAC,UAAU,CAAC,UAAgD,CAAC;YACvF,MAAM,GAAG,GAAG,SAAS,CAAC,UAAU,CAAC,GAA2B,CAAC;YAC7D,MAAM,aAAa,GAAG,SAAS,CAAC,UAAU,CAAC,aAAmC,CAAC;YAE/E,IAAI,QAAQ,IAAI,GAAG,IAAI,aAAa,KAAK,SAAS,EAAE,CAAC;gBACnD,MAAM,UAAU,GAAG,IAAI,GAAG,EAAkB,CAAC;gBAC7C,KAAK,MAAM,CAAC,IAAI,EAAE,GAAG,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,QAAQ,CAAC,EAAE,CAAC;oBACnD,UAAU,CAAC,GAAG,CAAC,IAAI,EAAE,GAAG,CAAC,CAAC;gBAC5B,CAAC;gBACD,IAAI,CAAC,UAAU,GAAG,EAAE,UAAU,EAAE,GAAG,EAAE,aAAa,EAAE,CAAC;YACvD,CAAC;QACH,CAAC;QAED,sBAAsB;QACtB,MAAM,cAAc,GAAG,IAAI,CAAC,KAAK,CAAC,cAAc,CAAC,WAAW,CAAC,CAAC;QAC9D,KAAK,MAAM,IAAI,IAAI,cAAc,EAAE,CAAC;YAClC,IAAI,IAAI,CAAC,EAAE,CAAC,UAAU,CAAC,eAAe,CAAC,EAAE,CAAC;gBACxC,MAAM,KAAK,GAAG,IAAI,CAAC,UAAU,CAAC;gBAC9B,IAAI,KAAK,CAAC,OAAO,IAAI,KAAK,CAAC,OAAO,EAAE,CAAC;oBACnC,IAAI,CAAC,aAAa,CAAC,GAAG,CAAC,KAAK,CAAC,OAAiB,EAAE;wBAC9C,OAAO,EAAE,KAAK,CAAC,OAAiB;wBAChC,OAAO,EAAE,KAAK,CAAC,OAAmB;wBAClC,IAAI,EAAG,KAAK,CAAC,IAAe,IAAI,CAAC;wBACjC,SAAS,EAAG,KAAK,CAAC,SAAoB,IAAI,IAAI;wBAC9C,MAAM,EAAG,KAAK,CAAC,MAAiB,IAAI,KAAK;wBACzC,aAAa,EAAG,KAAK,CAAC,aAAwB,IAAI,CAAC;wBACnD,gBAAgB,EAAG,KAAK,CAAC,gBAA2B,IAAI,CAAC;qBAC1D,CAAC,CAAC;gBACL,CAAC;YACH,CAAC;QACH,CAAC;QAED,gBAAgB;QAChB,MAAM,YAAY,GAAG,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,eAAe,CAAC,CAAC;QACzD,IAAI,YAAY,EAAE,CAAC;YACjB,MAAM,MAAM,GAAG,YAAY,CAAC,UAAU,CAAC,QAAqC,CAAC;YAC7E,IAAI,KAAK,CAAC,OAAO,CAAC,MAAM,CAAC,EAAE,CAAC;gBAC1B,IAAI,CAAC,QAAQ,GAAG,MAAM,CAAC;YACzB,CAAC;QACH,CAAC;IACH,CAAC;IAED,iEAAiE;IAEzD,kBAAkB,CAAC,OAAe,EAAE,WAAmB;QAC7D,OAAO;YACL,OAAO;YACP,OAAO,EAAE,IAAI,KAAK,CAAC,WAAW,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC;YACvC,IAAI,EAAE,CAAC;YACP,SAAS,EAAE,IAAI;YACf,MAAM,EAAE,KAAK;YACb,aAAa,EAAE,CAAC;YAChB,gBAAgB,EAAE,CAAC;SACpB,CAAC;IACJ,CAAC;IAEO,gBAAgB,CAAC,KAAmB,EAAE,IAAY;QACxD,OAAO,KAAK,CAAC,OAAO,CAAC,MAAM,GAAG,IAAI,EAAE,CAAC;YACnC,KAAK,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;QACxB,CAAC;IACH,CAAC;IAEO,gBAAgB;QACtB,MAAM,QAAQ,GAAG,IAAI,GAAG,EAAU,CAAC;QACnC,sBAAsB;QACtB,KAAK,MAAM,OAAO,IAAI,IAAI,CAAC,aAAa,CAAC,IAAI,EAAE,EAAE,CAAC;YAChD,QAAQ,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;QACxB,CAAC;QACD,gBAAgB;QAChB,KAAK,MAAM,EAAE,IAAI,IAAI,CAAC,QAAQ,EAAE,CAAC;YAC/B,QAAQ,CAAC,GAAG,CAAC,EAAE,CAAC,OAAO,CAAC,CAAC;QAC3B,CAAC;QACD,2BAA2B;QAC3B,MAAM,YAAY,GAAG,IAAI,CAAC,KAAK,CAAC,cAAc,CAAC,SAAS,CAAC,CAAC;QAC1D,KAAK,MAAM,IAAI,IAAI,YAAY,EAAE,CAAC;YAChC,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;QAC3B,CAAC;QACD,OAAO,CAAC,GAAG,QAAQ,CAAC,CAAC;IACvB,CAAC;IAEO,yBAAyB;QAC/B,MAAM,SAAS,GAAG,IAAI,CAAC,KAAK,CAAC,cAAc,CAAC,eAAe,CAAC,CAAC;QAC7D,OAAO,SAAS;aACb,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAE,CAAC,CAAC,UAAU,CAAC,eAA0B,IAAI,CAAC,CAAC,KAAK,CAAC;aAC/D,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,IAAI,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;IAC/C,CAAC;CACF"}

package/dist/runtime/graph-self-evolve.d.ts

@@ -0,0 +1,92 @@
+import type { ExecutionGraph } from './execution-graph.js';
+export interface EvolvedRule {
+    id: string;
+    name: string;
+    description: string;
+    condition: {
+        phaseKind?: string;
+        confidenceBelow?: number;
+        confidenceAbove?: number;
+        tokenRatioAbove?: number;
+        budgetRemainingBelow?: number;
+        hasBlockers?: boolean;
+    };
+    action: {
+        upgradeModel?: boolean;
+        downgradeModel?: boolean;
+        skipPhase?: boolean;
+        addPhase?: string;
+    };
+    confidence: number;
+    evidenceCount: number;
+    successRate: number;
+}
+export interface RuleProposal {
+    rule: EvolvedRule;
+    backtestedImprovement: number;
+    confidence: number;
+    evidence: string;
+}
+export interface BacktestResult {
+    improvement: number;
+    appliedCount: number;
+    wouldHaveHelped: number;
+    wouldHaveHurt: number;
+    falsePositiveRate: number;
+}
+/**
+ * Analyzes historical execution traces to propose improved replanning rules,
+ * backtests them against historical data, and evolves scoring weights based
+ * on user feedback.
+ *
+ * Rules are persisted as 'heuristic' nodes in the execution graph with id
+ * prefix 'rule:'. Feedback adjusts confidence: accepted → +0.1, rejected → -0.15.
+ */
+export declare class RuleEvolver {
+    private graph;
+    constructor(graph: ExecutionGraph);
+    /**
+     * Analyze historical orchestration traces and propose new replanning rules.
+     *
+     * 1. Extract features from all orchestrations via BFS.
+     * 2. Find patterns in failures: low-confidence phases, high token usage, blockers.
+     * 3. Generate candidate rules from observed failure patterns.
+     * 4. Backtest each candidate against all historical traces.
+     * 5. Return proposals sorted by improvement, filtered by confidence > 0.3.
+     */
+    evolveRules(traceCount?: number): RuleProposal[];
+    /**
+     * Backtest a single rule against all historical orchestrations.
+     *
+     * For each orchestration, extracts phase features via BFS. If the rule
+     * condition matches any phase, simulates the action:
+     * - upgradeModel on failed orch → wouldHaveHelped (optimistic)
+     * - upgradeModel on successful orch → wouldHaveHurt (false positive)
+     * - skipPhase on failed orch → wouldHaveHelped (conservative: saves tokens)
+     * - downgradeModel on successful orch → wouldHaveHelped (saves cost)
+     * - downgradeModel on failed orch → wouldHaveHurt
+     */
+    backtestRule(rule: EvolvedRule): BacktestResult;
+    /**
+     * Record user accept/reject of a rule proposal.
+     * Accepted → confidence +0.1 (capped at 1.0)
+     * Rejected → confidence -0.15 (floored at 0.0)
+     */
+    recordFeedback(ruleId: string, accepted: boolean): void;
+    /**
+     * Get all evolved rules currently stored in the graph.
+     * Loads heuristic nodes with id prefix 'rule:' and deserializes them.
+     */
+    getActiveRules(): EvolvedRule[];
+    private extractFeatures;
+    private inferPhaseConfidence;
+    private inferPhaseTokens;
+    private inferPhaseKind;
+    private generateCandidateRules;
+    private averageTokens;
+    private ruleMatchesOrchestration;
+    private saveRule;
+    private loadRules;
+    private formatEvidence;
+}
+//# sourceMappingURL=graph-self-evolve.d.ts.map

package/dist/runtime/graph-self-evolve.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"graph-self-evolve.d.ts","sourceRoot":"","sources":["../../src/runtime/graph-self-evolve.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AAK3D,MAAM,WAAW,WAAW;IAC1B,EAAE,EAAE,MAAM,CAAC;IACX,IAAI,EAAE,MAAM,CAAC;IACb,WAAW,EAAE,MAAM,CAAC;IACpB,SAAS,EAAE;QACT,SAAS,CAAC,EAAE,MAAM,CAAC;QACnB,eAAe,CAAC,EAAE,MAAM,CAAC;QACzB,eAAe,CAAC,EAAE,MAAM,CAAC;QACzB,eAAe,CAAC,EAAE,MAAM,CAAC;QACzB,oBAAoB,CAAC,EAAE,MAAM,CAAC;QAC9B,WAAW,CAAC,EAAE,OAAO,CAAC;KACvB,CAAC;IACF,MAAM,EAAE;QACN,YAAY,CAAC,EAAE,OAAO,CAAC;QACvB,cAAc,CAAC,EAAE,OAAO,CAAC;QACzB,SAAS,CAAC,EAAE,OAAO,CAAC;QACpB,QAAQ,CAAC,EAAE,MAAM,CAAC;KACnB,CAAC;IACF,UAAU,EAAE,MAAM,CAAC;IACnB,aAAa,EAAE,MAAM,CAAC;IACtB,WAAW,EAAE,MAAM,CAAC;CACrB;AAED,MAAM,WAAW,YAAY;IAC3B,IAAI,EAAE,WAAW,CAAC;IAClB,qBAAqB,EAAE,MAAM,CAAC;IAC9B,UAAU,EAAE,MAAM,CAAC;IACnB,QAAQ,EAAE,MAAM,CAAC;CAClB;AAED,MAAM,WAAW,cAAc;IAC7B,WAAW,EAAE,MAAM,CAAC;IACpB,YAAY,EAAE,MAAM,CAAC;IACrB,eAAe,EAAE,MAAM,CAAC;IACxB,aAAa,EAAE,MAAM,CAAC;IACtB,iBAAiB,EAAE,MAAM,CAAC;CAC3B;AAwBD;;;;;;;GAOG;AACH,qBAAa,WAAW;IACV,OAAO,CAAC,KAAK;gBAAL,KAAK,EAAE,cAAc;IAEzC;;;;;;;;OAQG;IACH,WAAW,CAAC,UAAU,CAAC,EAAE,MAAM,GAAG,YAAY,EAAE;IA2BhD;;;;;;;;;;OAUG;IACH,YAAY,CAAC,IAAI,EAAE,WAAW,GAAG,cAAc;IAsC/C;;;;OAIG;IACH,cAAc,CAAC,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,OAAO,GAAG,IAAI;IAsBvD;;;OAGG;IACH,cAAc,IAAI,WAAW,EAAE;IAM/B,OAAO,CAAC,eAAe;IAkDvB,OAAO,CAAC,oBAAoB;IAW5B,OAAO,CAAC,gBAAgB;IAKxB,OAAO,CAAC,cAAc;IAatB,OAAO,CAAC,sBAAsB;IA+I9B,OAAO,CAAC,aAAa;IAWrB,OAAO,CAAC,wBAAwB;IA+ChC,OAAO,CAAC,QAAQ;IAgBhB,OAAO,CAAC,SAAS;IAyBjB,OAAO,CAAC,cAAc;CASvB"}