Reactor for professional or team use

The evaluation starts from your activity: where the system can add value, which capabilities are needed, and how fit is measured. Working as a team also requires a separate check of access and permission requirements; connecting several accounts is not, on its own, a system for managing team permissions.

You can pick a focused task: scanning, data gathering, trade review, position management or reporting. Define what goes into the task, what needs to come out, and what will count as an improvement.

Reactor can also work alongside existing tools, through supported connections, incoming signals and passing results out.

Reactor gathers information from tools connected to markets and accounts. In analysis it's important to distinguish between a live price, a snapshot, a closed candle and a candle still forming.

You can check a data point's source, the time it refers to, and the scope of the check. Availability varies between markets, assets and brokers.

The research process lets you separate the opportunity, the entry decision, execution and trade management.

You can collect occurrences of defined conditions, replay the movement after them with the candle tools, and examine patterns. Then you add a decision layer and test its contribution, and later the position management.

The same opportunity can be assessed with different instructions or different models. For example, a conservative approach, one that looks for a breakout, or one that prefers to wait for a pullback.

You can run the comparison as analysis only, record the decisions and check the results later. That's how you test whether the differences in approach actually help.

You can work with connected accounts and examine positions, data and actions in the context of each account. The workflow needs to account for the broker, the assets, trading hours and the authority defined.

You can look at using similar instructions across several accounts, adapted to each one's data and capabilities.

You can split the process into detection, mapping, decision, risk management, profit management and reporting Agents. For each role you define the information required, the output and the authority.

You can keep human approval at selected steps. The AI's judgment operates inside the frame defined for each role.

Webhooks can deliver signals into Reactor and trigger analysis or a follow-up action. You can also pass information to external systems through supported connections.

An MCP connection lets another AI environment use capabilities exposed through Reactor, according to the connection's permissions and controls.

Run records let you review the model that ran, the tool calls and their results, failures, run duration and AI cost. Alongside that, you review the actions in the account and trading performance.

The trial should include both normal cases and relevant edge cases: missing information, a tool failure, an order that didn't complete, or a breach of the process conditions.