Professional architect sharing high‑quality architecture insights. Topics include high‑availability, high‑performance, high‑stability architectures, big data, machine learning, Java, system and distributed architecture, AI, and practical large‑scale architecture case studies. Open to ideas‑driven architects who enjoy sharing and learning.
The article analyzes how the growing strength of large language models shifts engineering bottlenecks from model capabilities to system stability, introducing the concept of a "Harness" that integrates models into real‑world workflows through state management, constraints, feedback loops, and verification mechanisms.
The article analyzes Matt Van Horn’s workflow that replaces traditional IDEs with a planning‑first approach using markdown task files, voice input, parallel research agents, and a custom "last30days" engine, showing how AI can orchestrate research, planning, execution, and verification to boost productivity across code and non‑code tasks.
The article analyzes Anthropic’s Harness design for long‑running applications, detailing how it mitigates context anxiety and self‑evaluation bias through sprint contracts, rubric scoring, and a planner‑generator‑evaluator architecture, and evaluates its effectiveness across multiple versions.
If you can only glance at the OpenAI Codex repository once, focus on five key entry points—AGENTS.md, justfile, rust-ci.yml, the app‑server README with schema fixtures, and rust-release.yml—to quickly understand the team’s engineering decisions, risk management, and release cadence.
The article analyses the Memento‑Skills paper and its open‑source implementation, showing how a frozen large language model can continuously improve by treating skills as external memory, using a five‑step Observe‑Read‑Act‑Feedback‑Write loop, advanced routing, and modular architecture to achieve significant gains on GAIA and HLE benchmarks.
The article analyses OpenAI’s approach to embedding AI‑driven Skills into a repository’s workflow—using AGENTS.md, scripts, and GitHub Actions—to automate repetitive engineering actions, improve PR throughput, and accelerate open‑source maintenance while keeping clear boundaries between model reasoning and deterministic scripts.
The article explains that Prompt Cache is not merely a way to reduce token costs, but a fundamental mechanism that forces developers to redesign the context management of long‑running AI agents, turning caching considerations into core architectural decisions.
The article explains that most OpenClaw issues stem from mixing up five core layers—channel, account, agent, session, and memory—and provides a step‑by‑step troubleshooting guide to isolate the problem, verify each layer, and configure the system for reliable responses.
This article outlines a practical, step‑by‑step workflow for Claude Code that starts with defining acceptance criteria, correctly layering context, selecting the right execution channel, enforcing system‑level constraints, and actively managing long sessions, turning experimental AI agents into reliable engineering tools.
The article explains how coding agents dramatically lower implementation costs, turning the bottleneck in engineering, product, and design collaborations from building code to making high‑quality judgments, reviewing, and context hand‑offs.
