How sullipaupau/hermes-agent works
Hermes Agent is a multi-interface AI agent: people chat through its terminal UI/CLI or messaging platforms, the agent uses a configurable model and tools to act, and responses return through the initiating interface. The README also describes persistent learning and recall, scheduled jobs, delegated subagents, and multiple execution backends as product capabilities. The graph groups these major branches with the core interaction flow; edges cite README statements or source-index references. Some subsystem wiring is unsampled, so those relationships are left out rather than inferred.
Main entry points: conversation_loop.py Conversation runtime · model_tools.py Tool dispatch · run.py Messaging gateway · auxiliary_client.py Model routing
Components 20
User interfaces
- CLI
cli.py - Terminal UI
tui_gateway/server.py - TUI client
ui-tui/src/app/turnController.ts - Messaging gateway
gateway/run.py - Platform adapters
gateway/platforms/base.py
Agent runtime
- Conversation runtime
agent/conversation_loop.py - Model routing
agent/auxiliary_client.py - Model providers
providers/base.py - Tool dispatch
model_tools.py - Tool registry
tools/registry.py - Execution backends
tools/environments/base.py
Agent capabilities
- Memory and recall
agent/memory_manager.py - Learning skills
agent/skill_utils.py - Session search
tools/session_search_tool.py - Subagents
tools/delegate_tool.py - Scheduled jobs
cron/scheduler.py
Configuration and operations
- Configuration
hermes_cli/config.py - Agent state
hermes_state.py
Other components
- User
- Model APIs
Connections 20
- User → CLI: starts chat
README.md - CLI → Conversation runtime: starts conversation
README.md - User → Terminal UI: chats in
README.md - Terminal UI → TUI client: serves UI
tui_gateway/server.py - User → Platform adapters: sends message
README.md - Platform adapters → Messaging gateway: routes messages
README.md - Messaging gateway → Conversation runtime: runs conversation
README.md - Conversation runtime → Model routing: requests completion
- Model routing → Model providers: resolves backend
agent/auxiliary_client.py - Model providers → Model APIs: calls model API
- Conversation runtime → Tool dispatch: dispatches tools
README.md - Tool dispatch → Tool registry: queries definitions
model_tools.py - Tool dispatch → Execution backends: runs commands
README.md - Conversation runtime → Memory and recall: curates memory
README.md - Conversation runtime → Learning skills: learns skills
README.md - Conversation runtime → Session search: recalls sessions
README.md - Conversation runtime → Subagents: spawns subagents
README.md - Scheduled jobs → Messaging gateway: delivers reports
README.md - Configuration → Model routing: configures providers
agent/auxiliary_client.py - Memory and recall → Agent state: persists state