dawoodtouseef/j.ar.v.i.s.

Purpose and launch surfaces

This is a Windows-oriented Python desktop automation assistant (“JARVIS”) supporting voice, text, camera, screenshot, and system-status interactions. Primary launch paths are the desktop application (main.py, jarvis.py, run_jarvis.bat), with selectable GPT All Star coding and Open Interpreter modes. The UI is a PyQt-based desktop client rooted in gui/AssistantGUI.py and gui/Home.py.

Desktop interaction and background workers

The GUI coordinates user-facing screens, terminal access, settings, login, and integrations. Background objects handle agent execution, alerts, speech recognition, TTS, wake-word activation, and consciousness/monitoring behavior. Key boundaries:

• Presentation/workers: gui/AssistantGUI.py, gui/Objects/AgentWorker.py, gui/Objects/SpeechRecognition.py

• Voice lifecycle: gui/Objects/wakeword.py, gui/Objects/TTSWorker.py

• Embedded terminal subsystem: gui/terminal/terminal_manager.py, gui/terminal/virtual_filesystem.py

Assistant brain, agent routing, and memory

The core assistant layer turns requests and sensed context into delegated agent work. It defines conversational, planning, proactive, security, financial, Gmail, vision, and personal-assistant roles; shared agent state provides the control/data boundary between orchestration and specialized actions. Persistent conversational memory is isolated under core/memory/.

• Orchestration: core/brain.py, core/agents/state.py

• Agent roles: core/agents/ConversationalAgent.py, core/agents/planner_agent.py, core/agents/vision_agents.py

• Persistence: core/memory/memory_agent.py

Automation, sensing, and external capabilities

Tools encapsulate OS-facing and external integrations: system health/sensors, scheduling, routing, web/news access, calculator/charting, home automation, software management, and stop control. This creates a practical boundary: agents decide intent; tools perform side effects and return observations.

• Tool catalog and routing: core/tools/router.py, core/tools/standard_tools.py

• Monitoring and proactive signals: core/tools/health_monitor.py, core/tools/sensors.py, core/tools/consciousness.py

• Automation integrations: core/tools/scheduler.py, core/tools/home_assistant_tool.py, core/tools/news_scraper.py

Computer-use interpreter runtime

The bundled Open Interpreter-style runtime offers an LLM-driven tool loop for desktop control. Its core resolves LLM responses/tool calls, then invokes browser, filesystem, terminal, display, keyboard, mouse, clipboard, contacts, calendar, mail, and vision adapters. It can be used independently through a terminal interface.

• Control loop and LLM dispatch: interpreter/core/core.py, interpreter/core/llm/llm.py, interpreter/computer_use/loop.py

• Computer abstraction: interpreter/core/computer/computer.py, interpreter/core/computer/terminal/terminal.py

• CLI/profiles: interpreter/terminal_interface/terminal_interface.py, interpreter/terminal_interface/profiles/profiles.py

GPT All Star coding pipeline

A separate multi-agent software-delivery workflow converts a coding request into staged specification, system/UI design, development, QA, healing, execution, and deployment. Role-specific agents collaborate through chains/team state; tools enable files, shell commands, and document/index retrieval. components/coder/ and components/gpt_all_star/ are near-parallel implementations and should be treated as duplicated architectural surfaces unless consolidated.

• Entrypoints/orchestration: components/coder/main.py, components/coder/core/gpt_all_star.py

• Roles and collaboration: components/coder/core/agents/agents.py, components/coder/core/team.py

• Pipeline stages: components/coder/core/steps/steps.py, components/coder/core/steps/development/development.py, components/coder/core/steps/quality_assurance/quality_assurance.py

• Execution boundary: components/coder/core/execution/execution.py, components/coder/core/tools/shell_tool.py

Infrastructure and persistence

The application is Python 3.12+ and Windows 10/11-focused, using PyQt5 and OpenCV per project documentation. Audio providers include OpenAI and local Whisper STT plus multiple TTS implementations. The integration layer contains database models and Alembic migrations, indicating persisted users, preferences, command history, contacts, agents, tools, sensors, and environment variables.

• Audio provider boundary: jarvis_integration/audio/stt_providers/openai.py, jarvis_integration/audio/tts_providers/BaseTTS.py

• Data models/database: jarvis_integration/internals/db.py, jarvis_integration/models/command_history.py

• Schema evolution: alembic.ini, migrations/env.py