5d499c552b
Long-term memory plugin using Memvid SDK with: - Efficient compressed storage (.mv2 format) - Full conversation history preservation - Hybrid search (semantic + lexical) - RAG capabilities - PII protection for all session transcripts
127 lines
3.0 KiB
TypeScript
127 lines
3.0 KiB
TypeScript
import { Type, type Static } from "@sinclair/typebox";
|
|
import { TypeCompiler } from "@sinclair/typebox/compiler";
|
|
|
|
/**
|
|
* Memory categories for classification
|
|
*/
|
|
export const MEMORY_CATEGORIES = [
|
|
"preference",
|
|
"decision",
|
|
"fact",
|
|
"entity",
|
|
"instruction",
|
|
"context",
|
|
"other",
|
|
] as const;
|
|
|
|
export type MemoryCategory = (typeof MEMORY_CATEGORIES)[number];
|
|
|
|
/**
|
|
* Configuration schema for Memvid memory plugin (local SDK)
|
|
*/
|
|
export const memvidConfigSchema = Type.Object({
|
|
// Storage Configuration
|
|
memoryPath: Type.Optional(
|
|
Type.String({
|
|
description: "Path to the .mv2 memory file (default: ~/.clawdbot/memories/moltbot.mv2)",
|
|
}),
|
|
),
|
|
|
|
// OpenAI Configuration (for embeddings and RAG)
|
|
openaiApiKey: Type.Optional(
|
|
Type.String({
|
|
description: "OpenAI API key for embeddings (uses OPENAI_API_KEY env var if not set)",
|
|
}),
|
|
),
|
|
embeddingModel: Type.Optional(
|
|
Type.String({
|
|
description: "OpenAI embedding model",
|
|
default: "text-embedding-3-small",
|
|
}),
|
|
),
|
|
|
|
// Behavior Configuration
|
|
autoRecall: Type.Optional(
|
|
Type.Boolean({
|
|
description: "Auto-inject relevant memories before agent starts",
|
|
default: true,
|
|
}),
|
|
),
|
|
autoCapture: Type.Optional(
|
|
Type.Boolean({
|
|
description: "Auto-capture important info after agent ends",
|
|
default: true,
|
|
}),
|
|
),
|
|
|
|
// Search Configuration
|
|
topK: Type.Optional(
|
|
Type.Number({
|
|
description: "Number of results to return from memory search",
|
|
default: 5,
|
|
minimum: 1,
|
|
maximum: 50,
|
|
}),
|
|
),
|
|
snippetChars: Type.Optional(
|
|
Type.Number({
|
|
description: "Maximum characters per snippet",
|
|
default: 500,
|
|
minimum: 100,
|
|
maximum: 2000,
|
|
}),
|
|
),
|
|
minScore: Type.Optional(
|
|
Type.Number({
|
|
description: "Minimum similarity score (0-1)",
|
|
default: 0.3,
|
|
minimum: 0,
|
|
maximum: 1,
|
|
}),
|
|
),
|
|
|
|
// RAG Configuration
|
|
ragModel: Type.Optional(
|
|
Type.String({
|
|
description: "Model for RAG answers (gpt-4o, gpt-4o-mini, etc.)",
|
|
default: "gpt-4o-mini",
|
|
}),
|
|
),
|
|
|
|
// Security Configuration
|
|
maskPii: Type.Optional(
|
|
Type.Boolean({
|
|
description: "Mask PII (emails, SSNs, phone numbers, credit cards, API keys) in recalled memories",
|
|
default: true,
|
|
}),
|
|
),
|
|
});
|
|
|
|
export type MemvidConfig = Static<typeof memvidConfigSchema>;
|
|
|
|
const configCompiler = TypeCompiler.Compile(memvidConfigSchema);
|
|
|
|
/**
|
|
* Parse and validate config with defaults
|
|
*/
|
|
export function parseConfig(raw: unknown): MemvidConfig {
|
|
// Handle undefined/null config
|
|
const config = raw && typeof raw === "object" ? raw : {};
|
|
|
|
if (!configCompiler.Check(config)) {
|
|
const errors = [...configCompiler.Errors(config)];
|
|
throw new Error(`Invalid config: ${errors.map((e) => `${e.path}: ${e.message}`).join(", ")}`);
|
|
}
|
|
return {
|
|
embeddingModel: "text-embedding-3-small",
|
|
autoRecall: true,
|
|
autoCapture: true,
|
|
topK: 5,
|
|
snippetChars: 500,
|
|
minScore: 0.3,
|
|
ragModel: "gpt-4o-mini",
|
|
maskPii: true, // Default: ON for security
|
|
...config,
|
|
} as MemvidConfig;
|
|
}
|