AI Agents

Agent development frameworks increasingly expose similar vocabulary: tools, sessions, memory, handoffs, and streaming. That vocabulary hides substantial architectural differences. Defining tools for an OpenAI agent, embedding Claude Code through the Claude Agent SDK, and compiling a stateful LangGraph workflow place different parts of the system under framework control. Comparing them through a list of supported features misses the decision that will determine most of the integration work: which execution responsibilities the application keeps, and which it delegates. ...

Engineering Reliable Coding Agents A coding agent can make a locally correct change while the project around it becomes less coherent. Consider a three-session authentication refactor. The first session chooses RS256 and implements token validation. The second session cannot recover that decision, infers HS256 instead, and introduces a conflicting path. The third session resolves the conflict but also rewrites an unrelated password module. Each session can plausibly report success. Taken together, the repository has drifted. ...