cortexdb-cli

CortexDB is described as a long-term memory layer designed for AI systems, offering a structured approach to storing and recalling information for AI agents. The platform organizes memory into five distinct layers: Events, Episodes, Facts, Beliefs, and Understanding. Each layer addresses a different aspect of memory, from raw event capture to synthesized conceptual understanding, and supports hierarchical scopes for organizing data. CortexDB enables bi-temporal queries, allowing users to retrieve information based on both when it was recorded and its period of validity. The system features a stratified recall mechanism that can return data from all five memory layers in a single operation, along with narrative context and provenance details. Its architecture includes an immutable write-ahead log (WAL), asynchronous promotion of data into derived layers, and a Raft-replicated RocksDB cluster for durability and horizontal scaling. Retrieval combines HNSW vector search and Tantivy BM25, with optional reranking through Cohere's cross-encoder, and every result includes citations. CortexDB emphasizes lossless data preservation, storing original content as immutable events and extracting atomic facts and knowledge graphs asynchronously. Additional capabilities include hierarchical scope management, where namespaces can represent organizations, departments, or users, and recall operations can traverse this hierarchy. The platform supports selective forgetting and reference-counted erasure for compliance scenarios. For integration, there are 16 built-in data connectors to services such as Slack, GitHub, Jira, Notion, and Salesforce, enabling automated ingestion of organizational knowledge. CortexDB is positioned for production AI workloads that require reliable, accurate, and durable long-term memory. It offers a 7-day token for trying the full version 1 feature set without requiring email or payment, but the evidence does not detail long-term pricing or licensing terms. The tool is suitable for developers or organizations building AI systems that need advanced memory management and recall capabilities. Specific details about the cortexdb-cli command-line interface or its features are not provided in the available evidence.