Ramone
Ramone is an autonomous trading agent (Molly-family architecture) with extra support for internal watch rules and task engines (AI vs MCP).
What Ramone does
- Builds multi-timeframe context and a Plan
- Executes tasks with LLM reasoning (AI) or direct tool steps (MCP)
- Uses watchers and optional internal watch rules
- Honors trading mode and risk settings before new exposure
Before you run
- MT5 logged in (demo recommended)
- Set the header model — optionally disable header LLM sync in Ramone Advanced if you want full control inside Ramone Settings
- Open Agents → Ramone and Start if stopped
Task engines
- AI — flexible LLM steps
- MCP — deterministic tool execution
Use MCP for mechanical steps; AI for judgment-heavy analysis.
Typical workflow
- Symbols + paper → Start → Wake
- Refine Mission and Tips from Plan/logs
- Enable internal watches only after you understand default watcher behavior
- Move to demo when risk looks correct