Personal Bot
Solo users
Local assistant for portfolio analysis, alerts, and constrained strategy automation.
Core Components
Default bot profileLocal model optionSafety guard
Delivery: Bundled with solo runtime
AI Trading Agents
Production + ExpansionOpeniBank agent mode separates decision proposal from execution authority. Brains can be rule-based, ML, LLM, or hybrid, but all spend-capable actions pass through safety and commitment gates.
North Star
Agents can reason over portfolio and market context while never bypassing deterministic risk, permit, and commitment controls.
Production Pillars
- •Brain abstraction (Rule / ML / LLM / Hybrid)
- •ProposedAction model separated from execution authority
- •Guard layer (budget, rate, anomaly, kill switch)
- •Local-first LLM option for privacy-sensitive use cases
- •Fleet lifecycle and observability for multi-agent operations
Source Baseline
This documentation page is aligned with the following OpeniBank v4 planning files.
- 06-AGENT-MODE.md
- OPENIBANK_BOT_SPEC.md
- 02-SOLO-MODE.md
Service Model
Team Agent Fleet
Trading teams and small funds
Multiple strategy agents with shared control plane and centralized risk envelope.
Core Components
Fleet managerRole policiesShared telemetry
Delivery: Hosted or self-hosted fleet controller
Agent Marketplace
Network participants
Distributable agent templates and skills with versioned policy constraints.
Core Components
Publishing pipelineDiscoveryRevenue controls
Delivery: Network-mode extension
Deep-Dive Pages
Back to all systemsAgent and Brain Architecture
ProductionAgents are stateful services that consume context, produce proposed actions, and submit through guarded execution interfaces.
Define a modular architecture where reasoning engines can evolve without weakening execution guarantees.
4 production checkpointsOpen Topic
Safety Guard Layer
ProductionBudget, rate, anomaly, and kill-switch controls create a hard execution perimeter around all agents.
Prevent unsafe autonomous behavior without removing useful automation.
4 production checkpointsOpen Topic
Skill System and SDK Surface
Production TrackSkill modules extend agent capability while preserving guardrail contracts.
Allow extensibility for ecosystem growth without introducing unbounded execution vectors.
4 production checkpointsOpen Topic
Agent Lifecycle and Fleet Operations
Production TrackProduction operation requires explicit agent state transitions, health tracking, and intervention controls.
Operate multi-agent systems reliably with predictable rollout, rollback, and incident handling.
4 production checkpointsOpen Topic