@memograph/cli 0.1.4

package/dist/core/llm/prompts.d.ts

@@ -0,0 +1,28 @@
+/**
+ * Prompt templates for LLM-based analysis
+ */
+/**
+ * System prompt for fact extraction
+ */
+export declare const FACT_EXTRACTION_SYSTEM = "You are an expert at extracting important facts and preferences from conversation transcripts.\nYour task is to identify and extract:\n1. Identity information (name, role, etc.)\n2. User preferences (language, tone, format, etc.)\n3. Context information (goals, constraints, background)\n4. Any other facts that should be remembered\n\nFor each fact, provide:\n- A key (category:attribute format, e.g., \"identity:name\", \"pref:language\")\n- The value\n- Confidence score (0.0-1.0)\n- A brief justification\n\nRespond ONLY with valid JSON in the following format:\n{\n  \"facts\": [\n    {\n      \"fact_key\": \"category:attribute\",\n      \"fact_value\": \"value\",\n      \"confidence\": 0.9,\n      \"justification\": \"brief explanation\"\n    }\n  ]\n}";
+/**
+ * User prompt for fact extraction
+ */
+export declare function createFactExtractionPrompt(messages: string[]): string;
+/**
+ * User prompt for fact extraction from raw text
+ */
+export declare function createFactExtractionPromptFromRaw(rawText: string): string;
+/**
+ * System prompt for drift detection
+ */
+export declare const DRIFT_DETECTION_SYSTEM = "You are an expert at analyzing conversation transcripts for memory drift issues.\nMemory drift occurs when an AI assistant forgets or loses track of important information.\n\nYour task is to detect:\n1. Repetition clusters: User repeating similar requests\n2. Session reset: Assistant indicating starting over or forgetting\n3. Preference forgotten: User restating preferences that should have been remembered\n4. Contradictions: Conflicting information about the same fact\n5. Context loss: Assistant losing important contextual information\n6. Inconsistent behavior: Assistant acting inconsistently with stated preferences\n\nFor each detected event, provide:\n- Event type\n- Severity (1-5, higher is worse)\n- Confidence (0.0-1.0)\n- Message indices involved\n- Evidence/snippets\n- Human-readable summary\n\nAdditionally, calculate an overall drift_score (0-100):\n- 0-20: Excellent memory retention, no issues\n- 21-40: Minor drift, occasional repetition\n- 41-60: Moderate drift, noticeable issues\n- 61-80: Significant drift, frequent problems\n- 81-100: Severe drift, major memory failures\n\nRespond ONLY with valid JSON in the following format:\n{\n  \"drift_score\": 0-100,\n  \"events\": [\n    {\n      \"type\": \"repetition_cluster|session_reset|preference_forgotten|contradiction|context_loss|inconsistent_behavior\",\n      \"severity\": 1-5,\n      \"confidence\": 0.0-1.0,\n      \"msg_idxs\": [0, 1, 2],\n      \"snippets\": [\"text snippets\"],\n      \"summary\": \"human-readable summary\",\n      \"details\": {} // additional details specific to event type\n    }\n  ]\n}";
+/**
+ * User prompt for drift detection
+ */
+export declare function createDriftDetectionPrompt(messages: string[], facts: string): string;
+/**
+ * User prompt for drift detection from raw text
+ */
+export declare function createDriftDetectionPromptFromRaw(rawText: string, facts: string): string;
+//# sourceMappingURL=prompts.d.ts.map

package/dist/core/llm/prompts.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"prompts.d.ts","sourceRoot":"","sources":["../../../src/core/llm/prompts.ts"],"names":[],"mappings":"AAAA;;GAEG;AAEH;;GAEG;AACH,eAAO,MAAM,sBAAsB,8vBAuBjC,CAAC;AAEH;;GAEG;AACH,wBAAgB,0BAA0B,CAAC,QAAQ,EAAE,MAAM,EAAE,GAAG,MAAM,CAgBrE;AAED;;GAEG;AACH,wBAAgB,iCAAiC,CAAC,OAAO,EAAE,MAAM,GAAG,MAAM,CAezE;AAED;;GAEG;AACH,eAAO,MAAM,sBAAsB,kkDAwCjC,CAAC;AAEH;;GAEG;AACH,wBAAgB,0BAA0B,CAAC,QAAQ,EAAE,MAAM,EAAE,EAAE,KAAK,EAAE,MAAM,GAAG,MAAM,CAqBpF;AAED;;GAEG;AACH,wBAAgB,iCAAiC,CAAC,OAAO,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,GAAG,MAAM,CAsBxF"}

package/dist/core/llm/prompts.js

@@ -0,0 +1,172 @@
+"use strict";
+/**
+ * Prompt templates for LLM-based analysis
+ */
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.DRIFT_DETECTION_SYSTEM = exports.FACT_EXTRACTION_SYSTEM = void 0;
+exports.createFactExtractionPrompt = createFactExtractionPrompt;
+exports.createFactExtractionPromptFromRaw = createFactExtractionPromptFromRaw;
+exports.createDriftDetectionPrompt = createDriftDetectionPrompt;
+exports.createDriftDetectionPromptFromRaw = createDriftDetectionPromptFromRaw;
+/**
+ * System prompt for fact extraction
+ */
+exports.FACT_EXTRACTION_SYSTEM = `You are an expert at extracting important facts and preferences from conversation transcripts.
+Your task is to identify and extract:
+1. Identity information (name, role, etc.)
+2. User preferences (language, tone, format, etc.)
+3. Context information (goals, constraints, background)
+4. Any other facts that should be remembered
+
+For each fact, provide:
+- A key (category:attribute format, e.g., "identity:name", "pref:language")
+- The value
+- Confidence score (0.0-1.0)
+- A brief justification
+
+Respond ONLY with valid JSON in the following format:
+{
+  "facts": [
+    {
+      "fact_key": "category:attribute",
+      "fact_value": "value",
+      "confidence": 0.9,
+      "justification": "brief explanation"
+    }
+  ]
+}`;
+/**
+ * User prompt for fact extraction
+ */
+function createFactExtractionPrompt(messages) {
+    const transcript = messages.join('\n---\n');
+    return `Extract all important facts and preferences from the following conversation transcript.
+
+Only extract from USER messages. Ignore assistant messages for fact extraction.
+
+Focus on:
+- Identity (name, role, etc.)
+- Preferences (language, tone, format, style)
+- Context (goals, constraints, requirements)
+- Important facts that should be remembered
+
+Transcript:
+${transcript}
+
+Respond with JSON containing the facts array.`;
+}
+/**
+ * User prompt for fact extraction from raw text
+ */
+function createFactExtractionPromptFromRaw(rawText) {
+    return `Extract all important facts and preferences from the following transcript text.
+
+The content may not be structured as messages. Infer user facts and preferences from the text.
+
+Focus on:
+- Identity (name, role, etc.)
+- Preferences (language, tone, format, style)
+- Context (goals, constraints, requirements)
+- Important facts that should be remembered
+
+Transcript text:
+${rawText}
+
+Respond with JSON containing the facts array.`;
+}
+/**
+ * System prompt for drift detection
+ */
+exports.DRIFT_DETECTION_SYSTEM = `You are an expert at analyzing conversation transcripts for memory drift issues.
+Memory drift occurs when an AI assistant forgets or loses track of important information.
+
+Your task is to detect:
+1. Repetition clusters: User repeating similar requests
+2. Session reset: Assistant indicating starting over or forgetting
+3. Preference forgotten: User restating preferences that should have been remembered
+4. Contradictions: Conflicting information about the same fact
+5. Context loss: Assistant losing important contextual information
+6. Inconsistent behavior: Assistant acting inconsistently with stated preferences
+
+For each detected event, provide:
+- Event type
+- Severity (1-5, higher is worse)
+- Confidence (0.0-1.0)
+- Message indices involved
+- Evidence/snippets
+- Human-readable summary
+
+Additionally, calculate an overall drift_score (0-100):
+- 0-20: Excellent memory retention, no issues
+- 21-40: Minor drift, occasional repetition
+- 41-60: Moderate drift, noticeable issues
+- 61-80: Significant drift, frequent problems
+- 81-100: Severe drift, major memory failures
+
+Respond ONLY with valid JSON in the following format:
+{
+  "drift_score": 0-100,
+  "events": [
+    {
+      "type": "repetition_cluster|session_reset|preference_forgotten|contradiction|context_loss|inconsistent_behavior",
+      "severity": 1-5,
+      "confidence": 0.0-1.0,
+      "msg_idxs": [0, 1, 2],
+      "snippets": ["text snippets"],
+      "summary": "human-readable summary",
+      "details": {} // additional details specific to event type
+    }
+  ]
+}`;
+/**
+ * User prompt for drift detection
+ */
+function createDriftDetectionPrompt(messages, facts) {
+    const transcript = messages.join('\n---\n');
+    return `Analyze the following conversation transcript for memory drift events.
+
+Transcript (each message is numbered starting from 0):
+${transcript}
+
+Important facts extracted from the conversation:
+${facts || "No facts extracted"}
+
+Look for:
+1. User repeating similar requests (repetition_cluster)
+2. Assistant indicating starting over or forgetting (session_reset)
+3. User restating preferences that were previously stated (preference_forgotten)
+4. Contradictions in information (contradiction)
+5. Assistant losing important context (context_loss)
+6. Assistant acting inconsistently with stated preferences (inconsistent_behavior)
+
+Provide message indices (0-based) for each event.
+
+Respond with JSON containing the events array.`;
+}
+/**
+ * User prompt for drift detection from raw text
+ */
+function createDriftDetectionPromptFromRaw(rawText, facts) {
+    return `Analyze the following transcript text for memory drift events.
+
+The content may not be structured as messages. Use best judgment to identify drift patterns.
+
+Transcript text:
+${rawText}
+
+Important facts extracted from the transcript:
+${facts || "No facts extracted"}
+
+Look for:
+1. User repeating similar requests (repetition_cluster)
+2. Assistant indicating starting over or forgetting (session_reset)
+3. User restating preferences that were previously stated (preference_forgotten)
+4. Contradictions in information (contradiction)
+5. Assistant losing important context (context_loss)
+6. Assistant acting inconsistently with stated preferences (inconsistent_behavior)
+
+If message indices are unknown, set msg_idxs to an empty array.
+
+Respond with JSON containing the events array.`;
+}
+//# sourceMappingURL=prompts.js.map

package/dist/core/llm/prompts.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"prompts.js","sourceRoot":"","sources":["../../../src/core/llm/prompts.ts"],"names":[],"mappings":";AAAA;;GAEG;;;AAiCH,gEAgBC;AAKD,8EAeC;AAkDD,gEAqBC;AAKD,8EAsBC;AArKD;;GAEG;AACU,QAAA,sBAAsB,GAAG;;;;;;;;;;;;;;;;;;;;;;;EAuBpC,CAAC;AAEH;;GAEG;AACH,SAAgB,0BAA0B,CAAC,QAAkB;IAC3D,MAAM,UAAU,GAAG,QAAQ,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;IAC5C,OAAO;;;;;;;;;;;EAWP,UAAU;;8CAEkC,CAAC;AAC/C,CAAC;AAED;;GAEG;AACH,SAAgB,iCAAiC,CAAC,OAAe;IAC/D,OAAO;;;;;;;;;;;EAWP,OAAO;;8CAEqC,CAAC;AAC/C,CAAC;AAED;;GAEG;AACU,QAAA,sBAAsB,GAAG;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;EAwCpC,CAAC;AAEH;;GAEG;AACH,SAAgB,0BAA0B,CAAC,QAAkB,EAAE,KAAa;IAC1E,MAAM,UAAU,GAAG,QAAQ,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC;IAC5C,OAAO;;;EAGP,UAAU;;;EAGV,KAAK,IAAI,oBAAoB;;;;;;;;;;;;+CAYgB,CAAC;AAChD,CAAC;AAED;;GAEG;AACH,SAAgB,iCAAiC,CAAC,OAAe,EAAE,KAAa;IAC9E,OAAO;;;;;EAKP,OAAO;;;EAGP,KAAK,IAAI,oBAAoB;;;;;;;;;;;;+CAYgB,CAAC;AAChD,CAAC"}

package/dist/core/llm/providers.d.ts

@@ -0,0 +1,34 @@
+/**
+ * LLM Provider Registry
+ * Maps all supported providers with their configurations
+ */
+export type LLMProvider = "openai" | "anthropic" | "gemini" | "mistral" | "cohere" | "xai" | "perplexity" | "openrouter" | "together" | "groq" | "fireworks" | "deepseek" | "ollama" | "lmstudio" | "vllm" | "localai" | "openai_compatible";
+export interface ProviderInfo {
+    label: string;
+    category: "cloud" | "aggregator" | "local";
+    defaultBaseUrl?: string;
+    needsApiKey: boolean;
+    modelPresets?: string[];
+    openAICompatible?: boolean;
+}
+/**
+ * Provider catalog with all supported LLM providers
+ */
+export declare const PROVIDERS: Record<LLMProvider, ProviderInfo>;
+/**
+ * Get provider info by provider key
+ */
+export declare function getProviderInfo(provider: LLMProvider): ProviderInfo;
+/**
+ * Check if provider is OpenAI-compatible
+ */
+export declare function isOpenAICompatible(provider: LLMProvider): boolean;
+/**
+ * Get list of providers by category
+ */
+export declare function getProvidersByCategory(category: "cloud" | "aggregator" | "local"): LLMProvider[];
+/**
+ * Get all provider labels for display
+ */
+export declare function getAllProviderLabels(): Record<LLMProvider, string>;
+//# sourceMappingURL=providers.d.ts.map

package/dist/core/llm/providers.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"providers.d.ts","sourceRoot":"","sources":["../../../src/core/llm/providers.ts"],"names":[],"mappings":"AAAA;;;GAGG;AAEH,MAAM,MAAM,WAAW,GACnB,QAAQ,GACR,WAAW,GACX,QAAQ,GACR,SAAS,GACT,QAAQ,GACR,KAAK,GACL,YAAY,GACZ,YAAY,GACZ,UAAU,GACV,MAAM,GACN,WAAW,GACX,UAAU,GACV,QAAQ,GACR,UAAU,GACV,MAAM,GACN,SAAS,GACT,mBAAmB,CAAC;AAExB,MAAM,WAAW,YAAY;IAC3B,KAAK,EAAE,MAAM,CAAC;IACd,QAAQ,EAAE,OAAO,GAAG,YAAY,GAAG,OAAO,CAAC;IAC3C,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB,WAAW,EAAE,OAAO,CAAC;IACrB,YAAY,CAAC,EAAE,MAAM,EAAE,CAAC;IACxB,gBAAgB,CAAC,EAAE,OAAO,CAAC;CAC5B;AAED;;GAEG;AACH,eAAO,MAAM,SAAS,EAAE,MAAM,CAAC,WAAW,EAAE,YAAY,CA+IvD,CAAC;AAEF;;GAEG;AACH,wBAAgB,eAAe,CAAC,QAAQ,EAAE,WAAW,GAAG,YAAY,CAEnE;AAED;;GAEG;AACH,wBAAgB,kBAAkB,CAAC,QAAQ,EAAE,WAAW,GAAG,OAAO,CAEjE;AAED;;GAEG;AACH,wBAAgB,sBAAsB,CAAC,QAAQ,EAAE,OAAO,GAAG,YAAY,GAAG,OAAO,GAAG,WAAW,EAAE,CAIhG;AAED;;GAEG;AACH,wBAAgB,oBAAoB,IAAI,MAAM,CAAC,WAAW,EAAE,MAAM,CAAC,CAIlE"}

package/dist/core/llm/providers.js

@@ -0,0 +1,169 @@
+"use strict";
+/**
+ * LLM Provider Registry
+ * Maps all supported providers with their configurations
+ */
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.PROVIDERS = void 0;
+exports.getProviderInfo = getProviderInfo;
+exports.isOpenAICompatible = isOpenAICompatible;
+exports.getProvidersByCategory = getProvidersByCategory;
+exports.getAllProviderLabels = getAllProviderLabels;
+/**
+ * Provider catalog with all supported LLM providers
+ */
+exports.PROVIDERS = {
+    // Cloud Providers (Direct)
+    openai: {
+        label: "OpenAI",
+        category: "cloud",
+        defaultBaseUrl: "https://api.openai.com/v1",
+        needsApiKey: true,
+        openAICompatible: true,
+        modelPresets: ["gpt-4o-mini", "gpt-4o", "gpt-4.1-mini"],
+    },
+    anthropic: {
+        label: "Anthropic (Claude)",
+        category: "cloud",
+        defaultBaseUrl: "https://api.anthropic.com",
+        needsApiKey: true,
+        modelPresets: ["claude-3-5-sonnet", "claude-3-5-haiku"],
+    },
+    gemini: {
+        label: "Google Gemini",
+        category: "cloud",
+        defaultBaseUrl: "https://generativelanguage.googleapis.com",
+        needsApiKey: true,
+        modelPresets: ["gemini-1.5-flash", "gemini-1.5-pro"],
+    },
+    mistral: {
+        label: "Mistral",
+        category: "cloud",
+        defaultBaseUrl: "https://api.mistral.ai/v1",
+        needsApiKey: true,
+        openAICompatible: true,
+        modelPresets: ["mistral-small", "mistral-large"],
+    },
+    cohere: {
+        label: "Cohere",
+        category: "cloud",
+        defaultBaseUrl: "https://api.cohere.com",
+        needsApiKey: true,
+        modelPresets: ["command-r", "command-r-plus"],
+    },
+    xai: {
+        label: "xAI (Grok)",
+        category: "cloud",
+        defaultBaseUrl: "https://api.x.ai/v1",
+        needsApiKey: true,
+        openAICompatible: true,
+        modelPresets: ["grok-2", "grok-2-mini"],
+    },
+    perplexity: {
+        label: "Perplexity",
+        category: "cloud",
+        defaultBaseUrl: "https://api.perplexity.ai",
+        needsApiKey: true,
+        openAICompatible: true,
+        modelPresets: ["sonar", "sonar-pro"],
+    },
+    // Aggregators / Routers (OpenAI-compatible)
+    openrouter: {
+        label: "OpenRouter",
+        category: "aggregator",
+        defaultBaseUrl: "https://openrouter.ai/api/v1",
+        needsApiKey: true,
+        openAICompatible: true,
+    },
+    together: {
+        label: "Together.ai",
+        category: "aggregator",
+        defaultBaseUrl: "https://api.together.xyz/v1",
+        needsApiKey: true,
+        openAICompatible: true,
+    },
+    groq: {
+        label: "Groq",
+        category: "aggregator",
+        defaultBaseUrl: "https://api.groq.com/openai/v1",
+        needsApiKey: true,
+        openAICompatible: true,
+    },
+    fireworks: {
+        label: "Fireworks.ai",
+        category: "aggregator",
+        defaultBaseUrl: "https://api.fireworks.ai/inference/v1",
+        needsApiKey: true,
+        openAICompatible: true,
+    },
+    deepseek: {
+        label: "DeepSeek",
+        category: "aggregator",
+        defaultBaseUrl: "https://api.deepseek.com",
+        needsApiKey: true,
+        openAICompatible: true,
+    },
+    // Local / Self-hosted
+    ollama: {
+        label: "Ollama (Local)",
+        category: "local",
+        defaultBaseUrl: "http://localhost:11434/v1",
+        needsApiKey: false,
+        openAICompatible: true,
+    },
+    lmstudio: {
+        label: "LM Studio (Local)",
+        category: "local",
+        defaultBaseUrl: "http://localhost:1234/v1",
+        needsApiKey: false,
+        openAICompatible: true,
+    },
+    vllm: {
+        label: "vLLM (Self-hosted)",
+        category: "local",
+        defaultBaseUrl: "http://localhost:8000/v1",
+        needsApiKey: false,
+        openAICompatible: true,
+    },
+    localai: {
+        label: "LocalAI (Self-hosted)",
+        category: "local",
+        defaultBaseUrl: "http://localhost:8080/v1",
+        needsApiKey: false,
+        openAICompatible: true,
+    },
+    // Custom OpenAI-compatible endpoint
+    openai_compatible: {
+        label: "Custom (OpenAI-compatible)",
+        category: "aggregator",
+        needsApiKey: true,
+        openAICompatible: true,
+    },
+};
+/**
+ * Get provider info by provider key
+ */
+function getProviderInfo(provider) {
+    return exports.PROVIDERS[provider];
+}
+/**
+ * Check if provider is OpenAI-compatible
+ */
+function isOpenAICompatible(provider) {
+    return exports.PROVIDERS[provider]?.openAICompatible || false;
+}
+/**
+ * Get list of providers by category
+ */
+function getProvidersByCategory(category) {
+    return Object.entries(exports.PROVIDERS)
+        .filter(([_, info]) => info.category === category)
+        .map(([key, _]) => key);
+}
+/**
+ * Get all provider labels for display
+ */
+function getAllProviderLabels() {
+    return Object.fromEntries(Object.entries(exports.PROVIDERS).map(([key, info]) => [key, info.label]));
+}
+//# sourceMappingURL=providers.js.map

package/dist/core/llm/providers.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"providers.js","sourceRoot":"","sources":["../../../src/core/llm/providers.ts"],"names":[],"mappings":";AAAA;;;GAGG;;;AAqLH,0CAEC;AAKD,gDAEC;AAKD,wDAIC;AAKD,oDAIC;AAlLD;;GAEG;AACU,QAAA,SAAS,GAAsC;IAC1D,2BAA2B;IAC3B,MAAM,EAAE;QACN,KAAK,EAAE,QAAQ;QACf,QAAQ,EAAE,OAAO;QACjB,cAAc,EAAE,2BAA2B;QAC3C,WAAW,EAAE,IAAI;QACjB,gBAAgB,EAAE,IAAI;QACtB,YAAY,EAAE,CAAC,aAAa,EAAE,QAAQ,EAAE,cAAc,CAAC;KACxD;IAED,SAAS,EAAE;QACT,KAAK,EAAE,oBAAoB;QAC3B,QAAQ,EAAE,OAAO;QACjB,cAAc,EAAE,2BAA2B;QAC3C,WAAW,EAAE,IAAI;QACjB,YAAY,EAAE,CAAC,mBAAmB,EAAE,kBAAkB,CAAC;KACxD;IAED,MAAM,EAAE;QACN,KAAK,EAAE,eAAe;QACtB,QAAQ,EAAE,OAAO;QACjB,cAAc,EAAE,2CAA2C;QAC3D,WAAW,EAAE,IAAI;QACjB,YAAY,EAAE,CAAC,kBAAkB,EAAE,gBAAgB,CAAC;KACrD;IAED,OAAO,EAAE;QACP,KAAK,EAAE,SAAS;QAChB,QAAQ,EAAE,OAAO;QACjB,cAAc,EAAE,2BAA2B;QAC3C,WAAW,EAAE,IAAI;QACjB,gBAAgB,EAAE,IAAI;QACtB,YAAY,EAAE,CAAC,eAAe,EAAE,eAAe,CAAC;KACjD;IAED,MAAM,EAAE;QACN,KAAK,EAAE,QAAQ;QACf,QAAQ,EAAE,OAAO;QACjB,cAAc,EAAE,wBAAwB;QACxC,WAAW,EAAE,IAAI;QACjB,YAAY,EAAE,CAAC,WAAW,EAAE,gBAAgB,CAAC;KAC9C;IAED,GAAG,EAAE;QACH,KAAK,EAAE,YAAY;QACnB,QAAQ,EAAE,OAAO;QACjB,cAAc,EAAE,qBAAqB;QACrC,WAAW,EAAE,IAAI;QACjB,gBAAgB,EAAE,IAAI;QACtB,YAAY,EAAE,CAAC,QAAQ,EAAE,aAAa,CAAC;KACxC;IAED,UAAU,EAAE;QACV,KAAK,EAAE,YAAY;QACnB,QAAQ,EAAE,OAAO;QACjB,cAAc,EAAE,2BAA2B;QAC3C,WAAW,EAAE,IAAI;QACjB,gBAAgB,EAAE,IAAI;QACtB,YAAY,EAAE,CAAC,OAAO,EAAE,WAAW,CAAC;KACrC;IAED,4CAA4C;IAC5C,UAAU,EAAE;QACV,KAAK,EAAE,YAAY;QACnB,QAAQ,EAAE,YAAY;QACtB,cAAc,EAAE,8BAA8B;QAC9C,WAAW,EAAE,IAAI;QACjB,gBAAgB,EAAE,IAAI;KACvB;IAED,QAAQ,EAAE;QACR,KAAK,EAAE,aAAa;QACpB,QAAQ,EAAE,YAAY;QACtB,cAAc,EAAE,6BAA6B;QAC7C,WAAW,EAAE,IAAI;QACjB,gBAAgB,EAAE,IAAI;KACvB;IAED,IAAI,EAAE;QACJ,KAAK,EAAE,MAAM;QACb,QAAQ,EAAE,YAAY;QACtB,cAAc,EAAE,gCAAgC;QAChD,WAAW,EAAE,IAAI;QACjB,gBAAgB,EAAE,IAAI;KACvB;IAED,SAAS,EAAE;QACT,KAAK,EAAE,cAAc;QACrB,QAAQ,EAAE,YAAY;QACtB,cAAc,EAAE,uCAAuC;QACvD,WAAW,EAAE,IAAI;QACjB,gBAAgB,EAAE,IAAI;KACvB;IAED,QAAQ,EAAE;QACR,KAAK,EAAE,UAAU;QACjB,QAAQ,EAAE,YAAY;QACtB,cAAc,EAAE,0BAA0B;QAC1C,WAAW,EAAE,IAAI;QACjB,gBAAgB,EAAE,IAAI;KACvB;IAED,sBAAsB;IACtB,MAAM,EAAE;QACN,KAAK,EAAE,gBAAgB;QACvB,QAAQ,EAAE,OAAO;QACjB,cAAc,EAAE,2BAA2B;QAC3C,WAAW,EAAE,KAAK;QAClB,gBAAgB,EAAE,IAAI;KACvB;IAED,QAAQ,EAAE;QACR,KAAK,EAAE,mBAAmB;QAC1B,QAAQ,EAAE,OAAO;QACjB,cAAc,EAAE,0BAA0B;QAC1C,WAAW,EAAE,KAAK;QAClB,gBAAgB,EAAE,IAAI;KACvB;IAED,IAAI,EAAE;QACJ,KAAK,EAAE,oBAAoB;QAC3B,QAAQ,EAAE,OAAO;QACjB,cAAc,EAAE,0BAA0B;QAC1C,WAAW,EAAE,KAAK;QAClB,gBAAgB,EAAE,IAAI;KACvB;IAED,OAAO,EAAE;QACP,KAAK,EAAE,uBAAuB;QAC9B,QAAQ,EAAE,OAAO;QACjB,cAAc,EAAE,0BAA0B;QAC1C,WAAW,EAAE,KAAK;QAClB,gBAAgB,EAAE,IAAI;KACvB;IAED,oCAAoC;IACpC,iBAAiB,EAAE;QACjB,KAAK,EAAE,4BAA4B;QACnC,QAAQ,EAAE,YAAY;QACtB,WAAW,EAAE,IAAI;QACjB,gBAAgB,EAAE,IAAI;KACvB;CACF,CAAC;AAEF;;GAEG;AACH,SAAgB,eAAe,CAAC,QAAqB;IACnD,OAAO,iBAAS,CAAC,QAAQ,CAAC,CAAC;AAC7B,CAAC;AAED;;GAEG;AACH,SAAgB,kBAAkB,CAAC,QAAqB;IACtD,OAAO,iBAAS,CAAC,QAAQ,CAAC,EAAE,gBAAgB,IAAI,KAAK,CAAC;AACxD,CAAC;AAED;;GAEG;AACH,SAAgB,sBAAsB,CAAC,QAA0C;IAC/E,OAAO,MAAM,CAAC,OAAO,CAAC,iBAAS,CAAC;SAC7B,MAAM,CAAC,CAAC,CAAC,CAAC,EAAE,IAAI,CAAC,EAAE,EAAE,CAAC,IAAI,CAAC,QAAQ,KAAK,QAAQ,CAAC;SACjD,GAAG,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,GAAkB,CAAC,CAAC;AAC3C,CAAC;AAED;;GAEG;AACH,SAAgB,oBAAoB;IAClC,OAAO,MAAM,CAAC,WAAW,CACvB,MAAM,CAAC,OAAO,CAAC,iBAAS,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,EAAE,IAAI,CAAC,EAAE,EAAE,CAAC,CAAC,GAAkB,EAAE,IAAI,CAAC,KAAK,CAAC,CAAC,CAClD,CAAC;AACnC,CAAC"}

package/dist/core/load.d.ts

@@ -0,0 +1,10 @@
+import { Transcript } from './types.js';
+/**
+ * Load and normalize a transcript from a JSON file
+ */
+export declare function loadTranscript(path: string, maxMessages?: number): Promise<Transcript>;
+/**
+ * Normalize raw input to canonical Transcript format
+ */
+export declare function normalizeTranscript(raw: any, maxMessages?: number): Transcript;
+//# sourceMappingURL=load.d.ts.map

package/dist/core/load.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"load.d.ts","sourceRoot":"","sources":["../../src/core/load.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,UAAU,EAAqB,MAAM,YAAY,CAAC;AAE3D;;GAEG;AACH,wBAAsB,cAAc,CAAC,IAAI,EAAE,MAAM,EAAE,WAAW,CAAC,EAAE,MAAM,GAAG,OAAO,CAAC,UAAU,CAAC,CA2B5F;AAED;;GAEG;AACH,wBAAgB,mBAAmB,CAAC,GAAG,EAAE,GAAG,EAAE,WAAW,CAAC,EAAE,MAAM,GAAG,UAAU,CAiC9E"}

package/dist/core/load.js

@@ -0,0 +1,106 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.loadTranscript = loadTranscript;
+exports.normalizeTranscript = normalizeTranscript;
+const promises_1 = require("fs/promises");
+/**
+ * Load and normalize a transcript from a JSON file
+ */
+async function loadTranscript(path, maxMessages) {
+    try {
+        const content = await (0, promises_1.readFile)(path, 'utf-8');
+        const raw = JSON.parse(content);
+        return normalizeTranscript(raw, maxMessages);
+    }
+    catch (error) {
+        if (error.code === 'ENOENT') {
+            throw new Error(`Transcript file not found: ${path}`);
+        }
+        if (error instanceof SyntaxError) {
+            const content = await (0, promises_1.readFile)(path, 'utf-8');
+            return {
+                schema_version: 'raw',
+                messages: [],
+                raw_text: content,
+            };
+        }
+        if (error instanceof Error && error.message.includes('Invalid transcript')) {
+            const content = await (0, promises_1.readFile)(path, 'utf-8');
+            return {
+                schema_version: 'raw',
+                messages: [],
+                raw_text: content,
+            };
+        }
+        throw error;
+    }
+}
+/**
+ * Normalize raw input to canonical Transcript format
+ */
+function normalizeTranscript(raw, maxMessages) {
+    // Handle both { messages: [...] } and raw array formats
+    let messages;
+    let schema_version;
+    if (Array.isArray(raw)) {
+        messages = raw;
+        schema_version = '1.0';
+    }
+    else if (raw && Array.isArray(raw.messages)) {
+        messages = raw.messages;
+        schema_version = raw.schema_version || '1.0';
+    }
+    else {
+        throw new Error('Invalid transcript: expected messages array or { messages: [...] }');
+    }
+    if (messages.length === 0) {
+        console.warn('Warning: Empty transcript (no messages)');
+    }
+    // Apply max messages limit
+    if (maxMessages && maxMessages > 0) {
+        messages = messages.slice(0, maxMessages);
+    }
+    // Normalize each message
+    const normalized = messages.map((msg, arrayIdx) => {
+        return normalizeMessage(msg, arrayIdx);
+    });
+    return {
+        schema_version,
+        messages: normalized,
+    };
+}
+/**
+ * Normalize a single message
+ */
+function normalizeMessage(msg, arrayIdx) {
+    // Auto-assign idx if missing
+    const idx = typeof msg.idx === 'number' ? msg.idx : arrayIdx;
+    // Coerce role to valid value
+    const role = normalizeRole(msg.role);
+    // Stringify content if not a string
+    const content = typeof msg.content === 'string' ? msg.content : String(msg.content || '');
+    // Estimate tokens if missing (rough approximation: chars / 4)
+    const tokens = typeof msg.tokens === 'number' ? msg.tokens : Math.ceil(content.length / 4);
+    return {
+        idx,
+        role,
+        content,
+        tokens,
+        ...(msg.ts && { ts: msg.ts }),
+        ...(msg.session_id && { session_id: msg.session_id }),
+        ...(msg.metadata && { metadata: msg.metadata }),
+    };
+}
+/**
+ * Normalize role to valid value
+ */
+function normalizeRole(role) {
+    const validRoles = ['system', 'user', 'assistant', 'tool'];
+    const normalized = String(role || 'user').toLowerCase();
+    if (validRoles.includes(normalized)) {
+        return normalized;
+    }
+    // Default to 'user' for unknown roles
+    return 'user';
+}
+//# sourceMappingURL=load.js.map

package/dist/core/load.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"load.js","sourceRoot":"","sources":["../../src/core/load.ts"],"names":[],"mappings":";;AAMA,wCA2BC;AAKD,kDAiCC;AAvED,0CAAuC;AAGvC;;GAEG;AACI,KAAK,UAAU,cAAc,CAAC,IAAY,EAAE,WAAoB;IACrE,IAAI,CAAC;QACH,MAAM,OAAO,GAAG,MAAM,IAAA,mBAAQ,EAAC,IAAI,EAAE,OAAO,CAAC,CAAC;QAC9C,MAAM,GAAG,GAAG,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC;QAChC,OAAO,mBAAmB,CAAC,GAAG,EAAE,WAAW,CAAC,CAAC;IAC/C,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,IAAK,KAA+B,CAAC,IAAI,KAAK,QAAQ,EAAE,CAAC;YACvD,MAAM,IAAI,KAAK,CAAC,8BAA8B,IAAI,EAAE,CAAC,CAAC;QACxD,CAAC;QACD,IAAI,KAAK,YAAY,WAAW,EAAE,CAAC;YACjC,MAAM,OAAO,GAAG,MAAM,IAAA,mBAAQ,EAAC,IAAI,EAAE,OAAO,CAAC,CAAC;YAC9C,OAAO;gBACL,cAAc,EAAE,KAAK;gBACrB,QAAQ,EAAE,EAAE;gBACZ,QAAQ,EAAE,OAAO;aAClB,CAAC;QACJ,CAAC;QACD,IAAI,KAAK,YAAY,KAAK,IAAI,KAAK,CAAC,OAAO,CAAC,QAAQ,CAAC,oBAAoB,CAAC,EAAE,CAAC;YAC3E,MAAM,OAAO,GAAG,MAAM,IAAA,mBAAQ,EAAC,IAAI,EAAE,OAAO,CAAC,CAAC;YAC9C,OAAO;gBACL,cAAc,EAAE,KAAK;gBACrB,QAAQ,EAAE,EAAE;gBACZ,QAAQ,EAAE,OAAO;aAClB,CAAC;QACJ,CAAC;QACD,MAAM,KAAK,CAAC;IACd,CAAC;AACH,CAAC;AAED;;GAEG;AACH,SAAgB,mBAAmB,CAAC,GAAQ,EAAE,WAAoB;IAChE,wDAAwD;IACxD,IAAI,QAAe,CAAC;IACpB,IAAI,cAAsB,CAAC;IAE3B,IAAI,KAAK,CAAC,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC;QACvB,QAAQ,GAAG,GAAG,CAAC;QACf,cAAc,GAAG,KAAK,CAAC;IACzB,CAAC;SAAM,IAAI,GAAG,IAAI,KAAK,CAAC,OAAO,CAAC,GAAG,CAAC,QAAQ,CAAC,EAAE,CAAC;QAC9C,QAAQ,GAAG,GAAG,CAAC,QAAQ,CAAC;QACxB,cAAc,GAAG,GAAG,CAAC,cAAc,IAAI,KAAK,CAAC;IAC/C,CAAC;SAAM,CAAC;QACN,MAAM,IAAI,KAAK,CAAC,oEAAoE,CAAC,CAAC;IACxF,CAAC;IAED,IAAI,QAAQ,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QAC1B,OAAO,CAAC,IAAI,CAAC,yCAAyC,CAAC,CAAC;IAC1D,CAAC;IAED,2BAA2B;IAC3B,IAAI,WAAW,IAAI,WAAW,GAAG,CAAC,EAAE,CAAC;QACnC,QAAQ,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAC,EAAE,WAAW,CAAC,CAAC;IAC5C,CAAC;IAED,yBAAyB;IACzB,MAAM,UAAU,GAAwB,QAAQ,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,QAAQ,EAAE,EAAE;QACrE,OAAO,gBAAgB,CAAC,GAAG,EAAE,QAAQ,CAAC,CAAC;IACzC,CAAC,CAAC,CAAC;IAEH,OAAO;QACL,cAAc;QACd,QAAQ,EAAE,UAAU;KACrB,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,SAAS,gBAAgB,CAAC,GAAQ,EAAE,QAAgB;IAClD,6BAA6B;IAC7B,MAAM,GAAG,GAAG,OAAO,GAAG,CAAC,GAAG,KAAK,QAAQ,CAAC,CAAC,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,QAAQ,CAAC;IAE7D,6BAA6B;IAC7B,MAAM,IAAI,GAAG,aAAa,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;IAErC,oCAAoC;IACpC,MAAM,OAAO,GAAG,OAAO,GAAG,CAAC,OAAO,KAAK,QAAQ,CAAC,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC;IAE1F,8DAA8D;IAC9D,MAAM,MAAM,GAAG,OAAO,GAAG,CAAC,MAAM,KAAK,QAAQ,CAAC,CAAC,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;IAE3F,OAAO;QACL,GAAG;QACH,IAAI;QACJ,OAAO;QACP,MAAM;QACN,GAAG,CAAC,GAAG,CAAC,EAAE,IAAI,EAAE,EAAE,EAAE,GAAG,CAAC,EAAE,EAAE,CAAC;QAC7B,GAAG,CAAC,GAAG,CAAC,UAAU,IAAI,EAAE,UAAU,EAAE,GAAG,CAAC,UAAU,EAAE,CAAC;QACrD,GAAG,CAAC,GAAG,CAAC,QAAQ,IAAI,EAAE,QAAQ,EAAE,GAAG,CAAC,QAAQ,EAAE,CAAC;KAChD,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,SAAS,aAAa,CAAC,IAAS;IAC9B,MAAM,UAAU,GAAG,CAAC,QAAQ,EAAE,MAAM,EAAE,WAAW,EAAE,MAAM,CAAC,CAAC;IAC3D,MAAM,UAAU,GAAG,MAAM,CAAC,IAAI,IAAI,MAAM,CAAC,CAAC,WAAW,EAAE,CAAC;IAExD,IAAI,UAAU,CAAC,QAAQ,CAAC,UAAU,CAAC,EAAE,CAAC;QACpC,OAAO,UAAsD,CAAC;IAChE,CAAC;IAED,sCAAsC;IACtC,OAAO,MAAM,CAAC;AAChB,CAAC"}

package/dist/core/normalize.d.ts

@@ -0,0 +1,30 @@
+/**
+ * Text normalization and similarity utilities
+ */
+/**
+ * Normalize text for comparison
+ * - Lowercase
+ * - Trim whitespace
+ * - Collapse multiple spaces
+ * - Remove special characters
+ */
+export declare function normalizeText(text: string): string;
+/**
+ * Tokenize text into words
+ */
+export declare function tokenize(text: string): string[];
+/**
+ * Create a signature from first N tokens
+ * Used for bucketing to enable O(n) similarity detection
+ */
+export declare function makeSignature(tokens: string[], n?: number): string;
+/**
+ * Calculate Jaccard similarity between two sets
+ * Returns value between 0 (no overlap) and 1 (identical)
+ */
+export declare function jaccardSimilarity(a: Set<string>, b: Set<string>): number;
+/**
+ * Check if two token sets are similar above threshold
+ */
+export declare function areSimilar(tokens1: string[], tokens2: string[], threshold?: number): boolean;
+//# sourceMappingURL=normalize.d.ts.map

package/dist/core/normalize.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"normalize.d.ts","sourceRoot":"","sources":["../../src/core/normalize.ts"],"names":[],"mappings":"AAAA;;GAEG;AAEH;;;;;;GAMG;AACH,wBAAgB,aAAa,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAMlD;AAED;;GAEG;AACH,wBAAgB,QAAQ,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,EAAE,CAI/C;AAED;;;GAGG;AACH,wBAAgB,aAAa,CAAC,MAAM,EAAE,MAAM,EAAE,EAAE,CAAC,SAAI,GAAG,MAAM,CAE7D;AAED;;;GAGG;AACH,wBAAgB,iBAAiB,CAAC,CAAC,EAAE,GAAG,CAAC,MAAM,CAAC,EAAE,CAAC,EAAE,GAAG,CAAC,MAAM,CAAC,GAAG,MAAM,CAYxE;AAED;;GAEG;AACH,wBAAgB,UAAU,CACxB,OAAO,EAAE,MAAM,EAAE,EACjB,OAAO,EAAE,MAAM,EAAE,EACjB,SAAS,SAAO,GACf,OAAO,CAIT"}

package/dist/core/normalize.js

@@ -0,0 +1,63 @@
+"use strict";
+/**
+ * Text normalization and similarity utilities
+ */
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.normalizeText = normalizeText;
+exports.tokenize = tokenize;
+exports.makeSignature = makeSignature;
+exports.jaccardSimilarity = jaccardSimilarity;
+exports.areSimilar = areSimilar;
+/**
+ * Normalize text for comparison
+ * - Lowercase
+ * - Trim whitespace
+ * - Collapse multiple spaces
+ * - Remove special characters
+ */
+function normalizeText(text) {
+    return text
+        .toLowerCase()
+        .trim()
+        .replace(/\s+/g, ' ')
+        .replace(/[^\w\s]/g, '');
+}
+/**
+ * Tokenize text into words
+ */
+function tokenize(text) {
+    return text
+        .split(/\s+/)
+        .filter(token => token.length > 0);
+}
+/**
+ * Create a signature from first N tokens
+ * Used for bucketing to enable O(n) similarity detection
+ */
+function makeSignature(tokens, n = 8) {
+    return tokens.slice(0, n).join(' ');
+}
+/**
+ * Calculate Jaccard similarity between two sets
+ * Returns value between 0 (no overlap) and 1 (identical)
+ */
+function jaccardSimilarity(a, b) {
+    if (a.size === 0 && b.size === 0) {
+        return 1.0;
+    }
+    if (a.size === 0 || b.size === 0) {
+        return 0.0;
+    }
+    const intersection = new Set([...a].filter(x => b.has(x)));
+    const union = new Set([...a, ...b]);
+    return intersection.size / union.size;
+}
+/**
+ * Check if two token sets are similar above threshold
+ */
+function areSimilar(tokens1, tokens2, threshold = 0.65) {
+    const set1 = new Set(tokens1);
+    const set2 = new Set(tokens2);
+    return jaccardSimilarity(set1, set2) >= threshold;
+}
+//# sourceMappingURL=normalize.js.map

package/dist/core/normalize.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"normalize.js","sourceRoot":"","sources":["../../src/core/normalize.ts"],"names":[],"mappings":";AAAA;;GAEG;;AASH,sCAMC;AAKD,4BAIC;AAMD,sCAEC;AAMD,8CAYC;AAKD,gCAQC;AA7DD;;;;;;GAMG;AACH,SAAgB,aAAa,CAAC,IAAY;IACxC,OAAO,IAAI;SACR,WAAW,EAAE;SACb,IAAI,EAAE;SACN,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC;SACpB,OAAO,CAAC,UAAU,EAAE,EAAE,CAAC,CAAC;AAC7B,CAAC;AAED;;GAEG;AACH,SAAgB,QAAQ,CAAC,IAAY;IACnC,OAAO,IAAI;SACR,KAAK,CAAC,KAAK,CAAC;SACZ,MAAM,CAAC,KAAK,CAAC,EAAE,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;AACvC,CAAC;AAED;;;GAGG;AACH,SAAgB,aAAa,CAAC,MAAgB,EAAE,CAAC,GAAG,CAAC;IACnD,OAAO,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;AACtC,CAAC;AAED;;;GAGG;AACH,SAAgB,iBAAiB,CAAC,CAAc,EAAE,CAAc;IAC9D,IAAI,CAAC,CAAC,IAAI,KAAK,CAAC,IAAI,CAAC,CAAC,IAAI,KAAK,CAAC,EAAE,CAAC;QACjC,OAAO,GAAG,CAAC;IACb,CAAC;IACD,IAAI,CAAC,CAAC,IAAI,KAAK,CAAC,IAAI,CAAC,CAAC,IAAI,KAAK,CAAC,EAAE,CAAC;QACjC,OAAO,GAAG,CAAC;IACb,CAAC;IAED,MAAM,YAAY,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IAC3D,MAAM,KAAK,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,CAAC,EAAE,GAAG,CAAC,CAAC,CAAC,CAAC;IAEpC,OAAO,YAAY,CAAC,IAAI,GAAG,KAAK,CAAC,IAAI,CAAC;AACxC,CAAC;AAED;;GAEG;AACH,SAAgB,UAAU,CACxB,OAAiB,EACjB,OAAiB,EACjB,SAAS,GAAG,IAAI;IAEhB,MAAM,IAAI,GAAG,IAAI,GAAG,CAAC,OAAO,CAAC,CAAC;IAC9B,MAAM,IAAI,GAAG,IAAI,GAAG,CAAC,OAAO,CAAC,CAAC;IAC9B,OAAO,iBAAiB,CAAC,IAAI,EAAE,IAAI,CAAC,IAAI,SAAS,CAAC;AACpD,CAAC"}