Articles for people who
ship the stack
Rewrites on React, Node.js, TypeScript and AI. A translated body is shown when it exists. If it does not, you see the English text and a note that says so.
Tagged: harness-engineering
Agent Harness Engineering: State, Verification and Control Around the Model
A runtime view of production agents: why prompts and context aren't enough, and how durable state, enforcement, verification and observability close the gap.
13 min readThe Naked Agent: What Breaks When Nothing Guards the Model
Build the same unprotected travel agent in the Claude Agent SDK and LangChain Deep Agents, then watch three failures show which harness controls are missing.
9 min readHarness Engineering: The Environment That Makes AI Agents Reliable
Harness engineering is the environment around AI agents: six components, three pillars, how the term caught on in 2026, and nine principles for production.
6 min readThe Harness Is the Product: A Practical Guide to Harnesses in Agentic AI
Learn what an agent harness is and its seven parts, from tools and context to guardrails and evals, plus sub-agents, failure modes and frameworks worth knowing.
14 min readAI Agent Equals AI Model Plus Harness: What the Harness Does and Why
An agent is a model plus a harness. See how harnesses add tools, memory, loops and guardrails, how open-weight and open-source agents differ, and why evals matter.
10 min readHarness, Loop and Graph: Three Nested Layers of a Coding-Agent System
A failure symptom points to the layer to fix. How harness, loop and graph nest, where multi-agent helps, what adversarial checks can't promise, and a checklist.
15 min readPrompt, Context, Harness and Loop Engineering: The Four Layers of an AI Agent
Prompt engineering and context engineering cover two of four layers. What harness and loop engineering control, a GitHub triage example, and the build order.
12 min readWhy AI Agents Fail in Production: Harness Engineering Explained
Agents stall at 70% when the fix is the system around the model. A look at prompt, context and harness engineering, the six harness layers and generator-evaluator splits.
8 min readHarness Engineering: The Runtime That Turns an LLM Into an Agent
What actually makes a coding agent work: the loop, tool dispatch that survives bad calls, and allow/ask/deny permissions, shown with about 600 lines of plain Python.
13 min read
About these articles
Request a 24h estimate
Need the same stack in a production operator layer? Send the brief — estimate within 24 hours.