What does this skill do?

The Data Atlas Skill Agent Designer provides a clear framework for designing intelligent agents that monitor data sources, build market prediction portfolios, and draft parameter changes—while always delegating the final decision to a human. Its focus is on architectural and workflow planning, not on executing live operations.

Four-lane architecture
Organize the agent workflow into roles such as research collector, basket creator, risk reviewer, and human editor, each with clear responsibilities.
Multiple data sources
Define data sources such as APIs, social media platforms, and GitHub repositories, along with their access requirements and read-only keys.
Human Approval Points
It specifies specific points in the workflow where the user must approve, reject, or adjust proposed parameters before continuing.
Security and Audit Limits
Define prohibited actions, such as live trading execution, and configure audit logs for sensitive data.

Usage examples

🏗️ Complete architectural design
Design a market research agent with four roles: data collector, basket compiler, risk reviewer, and human editor for the prediction market.
📡 Font Specifications
Define the data sources (Itô API, X API, GitHub) and read-only access keys for a market analysis agent.
✅ Workflow with human approval
Create a workflow in which the agent proposes a basket of assets and a human must approve it before it is saved, with clear checkpoints.
🔒 Define non-goals
Specifies the prohibited objectives of a trading agent: prohibit live execution; prohibit the storage of private data without a contract.

Features

Four-lane architecture Structure with distinct roles: research collector, basket compiler, risk reviewer, and human editor.
Data Source Management List of data sources with access requirements, read-only keys, and configurable permission levels.
Human Approval Points Define specific points in the workflow where user intervention is required before proceeding.
Audit and Security Limits Records inputs, model outputs, sources, and the final human decision; defines objectives that are not explicitly permitted.
Implementable output Generate a ready-to-implement workflow specification that includes sources, gateways, roles, and non-goals.

Frequently asked questions

No. The skill is designed solely for planning architecture and workflow, not for executing operations, trading, or live actions.
Any API, social network, or repository (GitHub, Itô, X), as long as you define the access requirements and the necessary credentials.
Yes. The skill ensures that the final decision always rests with a human through configurable approval steps in the workflow.
No. The skill generates the flow specification; the subsequent implementation may require MCPs depending on the chosen data sources.

Related skills

More skills in the same category:

Data Atlas Agent Designer — Architecture of Investigation Agents with Claude AI

¿Prefieres escuchar el contenido? Genera la narración de audio con un clic.