How titechprabhasolutions/brahma-ai works
Brahma AI is a Windows-first personal assistant: a user speaks or enters a task, the Python application plans or dispatches work to local actions, and the Electron interface presents the interaction. Its documented capabilities include voice responses, multi-step automation, browser automation, and customizable UI. The graph separates the app, task execution, action domains, and the browser-agent service. Some subsystem links are based on README statements or SOURCE INDEX references; unsampled implementation details are not inferred.
Main entry points: executor.py Task executor · main.py Assistant runtime · server.js Browser agent service · agent.js Browser agent
Components 22
Assistant app
- Electron UI
electron/renderer/renderer.js - Assistant runtime
main.py - Voice pipeline
actions/voice_pipeline.py
Task execution
- Task planner
agent/planner.py - Task executor
agent/executor.py - Error recovery
agent/error_handler.py
Local actions
- File operations
actions/file_controller.py - Desktop control
actions/computer_control.py - Computer settings
actions/computer_settings.py - Browser control
actions/browser_control.py - Code and dev tools
actions/dev_agent.py - Communication tools
actions/send_message.py - Saved workflows
actions/workflow_manager.py
Browser automation
- Browser bridge
browser_agent_bridge.py - Browser agent service
browser-agent/server.js - Browser agent
browser-agent/agent.js - Browser actions
browser-agent/actions.js - Browser controller
browser-agent/browser.js - Browser session state
browser-agent/memory.js
User state
- User memory
memory/memory_manager.py
Other components
- User
- Gemini API
Connections 28
- User → Electron UI: enters requests
README.md - Electron UI → Assistant runtime: uses app
- User → Assistant runtime: speaks requests
README.md - Assistant runtime → Voice pipeline: uses voice
- Assistant runtime → Task planner: requests plans
main.py - Assistant runtime → Task executor: dispatches tasks
main.py - Assistant runtime → User memory: uses memory
main.py - Task planner → Gemini API: generates plans
- Task executor → Gemini API: requests text
- Error recovery → Gemini API: analyzes failures
agent/error_handler.py - Task executor → Error recovery: requests recovery
agent/executor.py - Task executor → File operations: dispatches tools
agent/executor.py - Task executor → Desktop control: dispatches tools
agent/executor.py - Task executor → Computer settings: dispatches tools
agent/executor.py - Task executor → Browser control: dispatches tools
agent/executor.py - Task executor → Code and dev tools: dispatches tools
agent/executor.py - Task executor → Communication tools: dispatches tools
agent/executor.py - Assistant runtime → Saved workflows: dispatches workflows
main.py - Saved workflows → Browser control: runs workflow steps
actions/workflow_manager.py - Saved workflows → Computer settings: runs workflow steps
actions/workflow_manager.py - Browser bridge → Browser agent service: exchanges commands
- Browser agent service → Browser agent: routes commands
browser-agent/server.js - Browser agent service → Browser controller: launches browser
browser-agent/server.js - Browser agent service → Browser actions: constructs action engine
browser-agent/server.js - Browser agent service → Browser session state: updates session state
browser-agent/server.js - Browser agent → Browser actions: executes actions
browser-agent/agent.js - Browser agent → Browser controller: reads active page
browser-agent/agent.js - Browser agent → Browser session state: stores snapshots
browser-agent/agent.js