About the maintainer
I have 15 years of programming experience, including 10 years primarily with Java in finance- and e-commerce-related engineering. For more than two years I have been practicing with agents, focusing on turning a business goal into steps that can run, be maintained, and be checked.
I use around US$2,000 worth of tokens per month on average, trying tools, studying workflow boundaries, and documenting reusable lessons in public resources. The site shares only a small part of that long-term practice.
Bring a concrete task, a personal workflow idea, or a business pilot question. Explore the resource catalog, trilingual guides, and on-demand index, or discuss how a familiar business task could become an Agent workflow.
Cross-industry practice in film production, investment research, and research prototypes. The site also shares Agent resources and working methods.
hekloShanghai@outlook.com
- 15 years in programming
- 10 years in Java, finance, and e-commerce
- More than 2 years with agents
- About US$2,000 in token use per month on average
Project work in practice
Start with a business challenge and turn familiar work into an Agent workflow that can run or be demonstrated.
Film production
Connecting the AI workflow for short-drama production
Production cycle: 30 days to 7 days
Business challenge
Short-drama production spans scriptwriting, animation design, and video production, with material often needing repeated organization as it moves between stages.
How it took shape
Connected an AI workflow across the existing process, from scriptwriting through animation design to video production, so teams could carry materials and content forward between stages.
Project result
The short-drama production cycle went from 30 days to 7 days, with less repeated organization between stages.
Personal investment workflows
Turning existing trading strategies into Agent workflows
AI-enabled within 10 days
Business challenge
Investors already have futures, stock, or crypto strategies and personal routines, while daily execution can still take substantial screen time.
How it took shape
Mapped existing strategies and personal routines, then organized suitable research and daily tasks into Agent workflows—shifting time from watching markets toward strategy research and life.
Project result
Several individual investors completed the AI-enabled workflow within 10 days, reducing time spent watching markets. The workflows were built around their existing strategies.
Research and hardware prototypes
A research-topic demo in one week
Demonstrable prototype in one week
Business challenge
Graduate students needed to turn a topic in bioscience, medical research, or robotics hardware into an initial result they could show and discuss.
How it took shape
Clarified the topic goal and available materials, then quickly built a demonstrable prototype that could continue to evolve.
Project result
Demos across these research areas were completed within a week, providing a starting point for discussion and iteration.
About Agent Wealth Hub
Turn agent technology into practical skills
This hub is designed first for finance and e-commerce operators, product teams, and business leads who know their work but are new to AI, as well as people starting from scratch with agents and personal workflows. Begin with a familiar task such as organizing material, preparing a brief, or mapping a repeated process. Follow the path of choosing a task, finding a tool, building a workflow, and checking the result to understand Agent skills, MCP tools, and automation. The resource center gathers source-linked skills, MCP, automation workflows, and quantitative learning material so you can decide what to try, how to verify it, and how to turn a useful approach into a tool that can be run and maintained. Whether you are setting up a personal workspace or improving one part of a team process, start with a goal, inputs, and checkable output.
Start with a goal and shape a skill set
Agent work often takes more than installing a single skill. Describe the task you want to complete, then use categories and search to find relevant skills, tools, and steps: organizing a collection of material, connecting a data source, turning repetitive work into a workflow, or adding evaluation and safety checks to an existing project. The 200 skill pages focus on identifiable capability components; the 200 SOPs break common goals into preparation, actions, verification, and troubleshooting. Try the smallest useful capability in your own agent environment, then add the MCP, permissions, and automation that the result calls for. Tools, data, and goals differ from person to person, so the catalog offers clear paths for choosing and assembling what you need.
A lightweight index built for agents
When an agent must read an entire knowledge base for every task, its context quickly fills with material it does not need. This hub separates discovery indexes, language lists, categories, pagination, and individual details. An agent can read a short index first, retrieve a page of summaries for a relevant topic, and then open only the skill or SOP details and source links needed. This on-demand retrieval reduces irrelevant context and retrieval-token use, keeps useful information available for the task, and makes compact JSON easy to cache, connect, and review. People can browse every page like a regular website; machine-readable indexes give agents another route to relevant material.
Follow each resource back to its source
Knowing where a tool comes from and how to check it matters as much as knowing what it can do. Skill pages link to an upstream repository, exact directory or file, and pinned version. SOP steps that rely on external guidance cite their sources too. Pages identify review or hands-on status, check dates, permissions, and known limitations so you can decide whether to try a step, read the documentation first, or check the project for updates. Follow the links to inspect the original guidance, versions, dependencies, and licenses, then verify a small scope in your own environment. Clear provenance makes material easier to review and helps developers bring useful practices into their projects.
Explore topics across the Agent ecosystem
The catalog covers agent setup and evaluation, developer tools, MCP servers and clients, integrations, memory and context, workflow orchestration, automation, document and data work, security, retrieval, research, writing, and quantitative learning for crypto assets. Follow one work area in depth or combine resources from several categories into a broader process. Pages show repository-level Stars, Forks, and observation time to help you understand project activity. When choosing a tool, also consider your goal, its license, maintenance, and how you can verify it.
Use small tasks to explore efficiency and opportunity
Bring agents into real work through tasks with clear boundaries and checkable results: cleaning structured data, organizing report material, maintaining documentation or a customer-support knowledge base, connecting a marketing content workflow, or making repeated research preparation reusable. Define inputs, outputs, and permissions first; test a small sample, record failures, and identify where human review is needed. Then turn a useful approach into a skill, playbook, or deliverable. Quantitative learning can follow a measured research path: understand data and assumptions, run historical experiments, examine bias and risk, then verify the process in simulation. Learning, automation, and creating useful work can help people find opportunities and apply their abilities. Practical work can begin with a small task, a clear goal, and a result you can review.
Take part in an open ecosystem
The Agent ecosystem grows as developers and users share tools, documentation, and experience. This hub freely organizes learning guidance for open Agent projects. When using upstream code, skills, documents, or media, follow the license and instructions of each project. If you find a stale link, a missing tool, or an unclear step, contact us through the public email; precise sources and suggestions help improve the catalog. If you want to adapt an idea to your own work, you can also explore custom skills, learning support, and automation workflow collaboration. These are separate paid services; scope, deliverables, and fees are agreed by email before work starts. Services are available in Chinese and English. Voluntary contributions support maintenance and are separate from content access and service purchases.