Skip to content

Latest commit

 

History

History
236 lines (183 loc) · 7.24 KB

File metadata and controls

236 lines (183 loc) · 7.24 KB

🤖 Agents Reference

eVera ships with 10 specialized agents, each designed for a specific domain. All agents extend BaseAgent and use the same tool calling framework.


💬 Companion Agent

File: vera/brain/agents/companion.py Tier: Executor (Ollama local) Description: Casual conversation, emotional support, jokes, mood checks.

Tool Description
chat Open-ended conversation
tell_joke Generate a joke
check_mood Empathetic mood check-in
suggest_activity Suggest fun activities based on mood

Example:

User: "Hey Vera, I'm feeling bored"
Vera: "I feel you! 😄 How about: 🎮 Try a new game, 📚 Start that book, 🎨 Do some creative sketching, or 🎵 Discover new music? What sounds fun?"

💻 Operator Agent

File: vera/brain/agents/operator.py Tier: Specialist (GPT-4o-mini) Description: Full PC automation — apps, scripts, files, mouse, keyboard, windows, processes, system info.

Tool Description
open_application Open apps by name (cross-platform maps)
execute_script Run shell/Python/PowerShell commands safely
manage_files Copy, move, delete, list, mkdir, info
take_screenshot Capture screen
type_text Type text into focused window
screen_capture Capture screen for AI analysis
analyze_screen AI vision analysis of screen content
ocr_screen OCR text extraction from screen
mouse_click Click at screen coordinates
mouse_move Move cursor to coordinates
mouse_drag Click and drag between points
scroll Scroll mouse wheel
press_hotkey Press keyboard shortcuts
manage_window List, focus, minimize, maximize, close windows
manage_process List, inspect, kill processes
system_info CPU, memory, disk, battery, OS details
manage_service List, start, stop, restart services
network_info IP, connections, ping
clipboard Read/write system clipboard
send_notification OS-level desktop notification

Example:

User: "Open Chrome and click at 500, 300"
Vera: "Done! ✅ Opening Chrome for you! 🚀"

🌐 Browser Agent

File: vera/brain/agents/browser.py Tier: Specialist (GPT-4o-mini) Description: Web automation via Playwright — navigate, login, fill forms, post on social media.

Tool Description
navigate Go to a URL
click Click an element on the page
fill_form Fill form fields
login Login to a website
type_in_browser Type text into browser elements
extract_text Extract text from page
page_screenshot Screenshot current page
analyze_page AI analysis of page content
get_page_elements List interactive elements
scroll Scroll the page
go_back Navigate back

Example:

User: "Go to twitter.com and post 'Hello world!'"
Vera: "I'll need to log in first. Should I go ahead? 🤔 (yes/no)"

🔍 Researcher Agent

File: vera/brain/agents/researcher.py Tier: Specialist (GPT-4o-mini) Description: Web search, URL summarization, academic papers, fact-checking.

Tool Description
web_search DuckDuckGo search (no API key needed)
summarize_url Summarize content from a URL
search_papers Search arXiv for academic papers
fact_check Verify claims against multiple sources

✍️ Writer Agent

File: vera/brain/agents/writer.py Tier: Specialist (GPT-4o-mini) Description: Draft, edit, format, and translate text.

Tool Description
draft_text Draft text content (email, blog, report)
edit_text Edit and improve existing text
format_document Format text (markdown, HTML)
translate Translate between languages

📅 Life Manager Agent

File: vera/brain/agents/life_manager.py Tier: Specialist (GPT-4o-mini) Description: Calendar, reminders, to-do lists, email.

Tool Description
check_calendar View upcoming events
add_event Create calendar event
create_reminder Set a reminder with time
list_todos / add_todo Manage to-do list
send_email Send email via SMTP

🏠 Home Controller Agent

File: vera/brain/agents/home_controller.py Tier: Executor (Ollama local) Description: Smart home device control via Home Assistant API.

Tool Description
control_light Turn lights on/off, set brightness/color
set_thermostat Set temperature
lock_door Lock/unlock doors
check_security Check security system status
play_media Control media playback
set_scene Activate predefined scenes

📈 Income Agent

File: vera/brain/agents/income.py + vera/brain/agents/brokers.py Tier: Strategist (GPT-4o) Description: Stock market monitoring, paper trading, and real broker integration.

Tool Description
get_stock_price Real-time stock quotes (yfinance)
get_stock_history Historical price data
view_portfolio View paper trading portfolio
paper_trade Buy/sell with virtual $100k
smart_trade Intelligent trade execution
alpaca_trade Alpaca broker integration
ibkr_trade Interactive Brokers integration
alpaca_account Alpaca account info
ibkr_account IBKR account info
market_overview Market summary
stock_news Latest stock news
watchlist Manage watchlist
tradingview_setup TradingView webhook config
automate_broker_app Broker app automation

💻 Coder Agent

File: vera/brain/agents/coder.py Tier: Specialist (GPT-4o-mini) Description: Read, write, edit code files and integrate with VS Code.

Tool Description
read_file Read file contents
write_file Create or overwrite files
edit_file Edit specific sections of files
search_in_files Grep/search across codebase
open_in_vscode Open file in VS Code

📦 Git Agent

File: vera/brain/agents/git_agent.py Tier: Specialist (GPT-4o-mini) Description: Git operations and AI-powered code review.

Tool Description
git_status Repository status
git_diff Show changes
git_commit Commit with message
git_push Push to remote
git_pull Pull from remote
git_branch Branch operations
git_log Commit history
git_stash Stash changes
git_review AI-powered code review

🔌 Plugin Agents

Directory: plugins/

Custom agents can be added as Python files in the plugins/ directory. They are auto-discovered and registered at startup. See plugins/_example_plugin.py for a template.


Multi-Agent Crews

File: vera/brain/crew.py

Multiple agents can collaborate on complex tasks using 4 strategies:

Strategy Description
sequential Agents process in order, passing results
parallel All agents work simultaneously
hierarchical Supervisor decomposes and delegates
debate Agents argue perspectives, best wins