Architecture and Systems Thinking
ESSENTIAL READING · START HERE
Read how decisions about ownership, authorization, evidence, failure attribution and deletion shape a long-lived agent. Some linked essays remain available only in Chinese.
Explore the nine core principles and six on-demand methods, including Context as a workbench, minimum sufficient mechanisms and replaceable capabilities. Existing essays are listed below; new and revised topics are clearly distinguished in the writing roadmap.
New principle essay: Local Improvements Are Not System Progress connects tool projection, tracing and evaluation to verifiable end-to-end outcomes without requiring expensive re-testing for every change.
Reliability case study: How Does an Agent Continue After a Model Failure? covers failure domains, original Run authorization and budgets, disclosure, and recovery identity. It describes controlled continuation, not completed seamless production high availability.
Research design lesson: A Research Agent Needs More Than Tools shows why model judgment, settled tool receipts, time-qualified financial evidence, optional subagents, and actual delivery belong to one evidence-driven research story. Also featured in Fintech × Agent.
Four reading paths, aligned with the October 25 roadmap: Deep and Continuous Research, Subagents and Experts, Finance and Engineering, and Other Architecture Topics. These are editorial subgroups within the existing architecture collection, not four new product systems or claims of production readiness. Module mechanisms and AI-assisted development remain in their separate collections.
01 / Architecture and Systems Thinking
System boundaries, responsibility, reliability and change.
Deep and Continuous Research
Long-running background research, staged delivery, recovery and continuity of investment theses.
Subagents and Experts
Investment perspectives, bounded peer dialogues and collaboration across expert agents.
Finance and Engineering
Financial evidence, reproducible results, system evolution and end-to-end engineering quality.
- Why Financial Market MCP Became a Decision System
- Boundaries Between Research, Advice and Trading
- Financial Data: Time, Definitions and Reproducibility
- Why Python Results Can Still Be Misused in Reports
- A Trustworthy Evidence Chain for Financial Agents (Chinese original)
- A Research Agent Needs More Than Tools: Model, Harness, and Loop
- Why More Citations Do Not Prove Reliability
- Building an ETF Research Assistant
- Code Is Cheaper; Why Are Systems Harder to Evolve?
- What Should Stay Stable—and What Should Change—in an Agent Architecture?
- Stop Blaming Every Failure on the Model
- Local Improvements Are Not System Progress
Other Architecture Topics
System overviews, model/runtime responsibilities, authorization and capability ownership.
Concurrent interaction: Keep Chatting — delegation, side questions, and adjustments in one conversation.