Deepractice/PromptX
PromptX
PromptX is an LLM agent orchestration framework designed to execute multi-step workflows using autonomous agents. It features a sandboxed tool execution environment for secure filesystem operations and external API integrations, alongside a persona management system that defines professional roles and domain expertise to control agent behavior.
The system implements a semantic memory network for persistent knowledge storage, utilizing graph-based memory and engrams to retain information across sessions. This cognitive memory includes specialized tools for knowledge graph visualization, allowing users to map and analyze memory networks through interactive mind maps and density statistics. To prevent knowledge contamination, the framework maintains isolated memory networks and workspaces for different professional personas.
The platform covers automated document processing for PDFs, spreadsheets, and Word documents, as well as a custom tool integration framework for connecting agents to external databases and services. It also includes utilities for exposing local servers via token-authenticated remote access URLs.
Features
- Semantic Memory Managers - Implements a persistent, graph-based semantic memory network using nodes and edges to simulate associative human memory.
- Agent Orchestration Frameworks - Provides a platform for deploying and managing autonomous agents with persistent memory and custom tool capabilities.
- Long-term Memory Stores - Implements persistent storage mechanisms to retain agent context and knowledge across multiple sessions.
- Autonomous Agents - Provides a framework for executing autonomous agent sessions with full tool access and isolated workspaces.
- Multi-Step Workflows - Sequences multiple tool calls automatically to transform raw input data into structured reports.
- Autonomous AI Agent Frameworks - Provides a framework for building autonomous agents that execute multi-step tasks and use tools in sandboxed environments.
- Cognitive Memory Systems - Simulates human-like memory and retrieval by storing and recalling concepts via semantic networks.
- Intelligent Document Processing - Extracts and analyzes data from PDFs, spreadsheets, and Word documents using intelligent caching.
- Persona Definitions - Provides named configurations and system prompts to define specialized AI personas and behavioral roles.
- System Instruction Personas - Injects structured identity profiles from configuration files into the system prompt to swap agent behavior.
- Multi-step Goal Execution - Chains multiple tool operations together automatically to transform raw data into formatted reports.
- Persona Management Systems - Offers a system for creating and organizing AI identities to control agent behavior and expertise.
- Knowledge Base Storage - Provides unified interfaces for persisting new information into a network to ensure it is retained across sessions.
- Sandboxed Execution Environments - Ships a secure runtime for executing custom tools and filesystem operations in isolated environments.
- Tool Execution Sandboxes - Executes tool operations and filesystem tasks within secure, isolated environments using parallel worker threads.
- Memory Isolation - Partitions agent memory spaces by persona to ensure data privacy and prevent knowledge contamination.
- AI Personas - Defines behavioral identities through reusable system prompts and configurations to control agent roles.
- Workspace Isolations - Maintains separate memory networks and file contexts for different professional personas to prevent knowledge contamination.
- Behavioral Role Switching - Allows agents to dynamically switch professional roles to adapt their behavior and expertise for specific tasks.
- Semantic Networks - Utilizes a persistent graph-based memory network and engrams to retain cognitive information.
- Memory Visualizers - Provides tools to visualize and edit memory networks and engrams to control knowledge storage.
- Contextual Information Retrieval - Implements techniques for fetching relevant context from memory networks using focused or creative search modes.
- Document Processing Toolsets - Integrates external capabilities to read and manipulate complex files like PDFs, spreadsheets, and Word documents.
- External Tool Integrations - Connects agents to external APIs and databases via Model Context Protocol to execute real-world tasks.
- Knowledge Mapping - Renders memory networks as interactive mind maps to provide a spatial representation of stored concepts.
- PDF Document Analyzers - Extracts text and images from PDF files using intelligent caching and on-demand page reading.
- Tool Integration Frameworks - Provides a standardized interface for equipping agents with custom wrappers to integrate external APIs.
- Word Document Management - Reads and writes structured Word documents while preserving original formatting and hierarchy.
- Spreadsheet Analysis Workflows - Processes Excel files to generate insights, automate reports, and create data visualizations.
- Strict Tool Argument Validators - Enforces strict parameter types and schema validation for external API and function calls before execution.
- On-Demand Document Caching - Optimizes analysis of large files by reading and caching specific pages or cells for reuse.
- Filesystem Operations - Performs filesystem read, write, and organization tasks using natural language commands and metadata inspection.
- Knowledge Graph Visualizations - Provides a visual interface for mapping semantic memory networks as networked nodes and edges.
