Harness design for long-running application development

kept by eddie mar 25

Anthropic engineer Prithvi Rajasekaran builds a multi-agent harness with planner, generator, and evaluator agents that produces full-stack applications over multi-hour autonomous coding sessions.

I built a vulnerable app and spent $1,500 seeing if LLMs could hack it
A security researcher built a deliberately vulnerable Firebase-based app and spent $1,500 testing whether LLMs could exploit it; GPT-5.5 succeeded 70% of the time, while most others failed due to guardrails or misdirected focus.
also on FastAPI, Anthropic, #ai-coding, #developer-tools
How to build a Frontend for LangChain Deep Agents with CopilotKit! | Blog
CopilotKit builds a real-time frontend for LangChain Deep Agents by streaming agent state to a Next.js UI via the AG-UI protocol and CopilotKit middleware.
also on FastAPI, #ai-agents, #developer-tools
Effective harnesses for long-running agents \ Anthropic
Anthropic developed a two-fold harness with an initializer agent and a coding agent to enable long-running agents to make incremental progress across multiple context windows.
also on Claude Agent SDK, Claude, Anthropic, #ai-agents, #ai-coding, #software-engineering, #developer-tools
The Software Factory: Why Your Team Will Never Work the Same Again
AI coding agents like Claude Code already enable software factories where developers design specs and agents implement, shrinking teams and shipping features in hours instead of weeks.
also on Claude Agent SDK, Claude Code, Anthropic, #ai-agents, #ai-coding, #software-engineering, #developer-tools
Agentic Engineering Patterns
Effective agentic coding requires a robust test harness, letting AI explore ideas freely, and maintaining a scratch pad of learnings between sessions.
also on Playwright, Claude, #ai-agents, #ai-coding, #software-engineering, #developer-tools