11 pages

sources

One page per ingested source — a faithful summary, key claims, notable quotes, and open questions.

Loop engineering: Getting started with loops

Delba de Oliveira, Michael Segner ingested Jul 12, 2026

Claude Code team's loop taxonomy: turn/goal/time-based + proactive; stop conditions.

为 Windows 上的 Codex 构建安全高效的沙箱环境

David Wiesen ingested Jul 12, 2026

OpenAI's Codex Windows sandbox: rejected primitives → unelevated prototype → elevated design.

Choosing a Claude model and effort level in Claude Code

Lydia Hallie ingested Jul 12, 2026

Anthropic's model-vs-effort explainer: "didn't know vs didn't try" triage.

Harness Engineering for Self-Improvement

Lilian Weng ingested Jul 7, 2026

Lilian Weng's Jul 2026 survey: harness engineering as the near-term path to RSI.

LLM Powered Autonomous Agents

Lilian Weng ingested Jun 4, 2026

Lilian Weng's 2023 foundational survey: agent = LLM brain + Planning + Memory + Tool use.

The Second Half

Shunyu Yao ingested Apr 27, 2026

Shunyu Yao's thesis: the recipe generalizes; evaluation is the bottleneck — the utility problem.

你不知道的大模型训练：原理、路径与新实践

Tw93 ingested Apr 27, 2026

Tw93's third essay: post-training, eval/reward, harness, distillation — the pipeline's back half.

Harness engineering: Leveraging Codex in an agent-first world

Ryan Lopopolo ingested Apr 27, 2026

Ryan Lopopolo / OpenAI Codex's "no manual code" 5-month, 1M-LoC, 1500-PR experiment.

Investigating how Codex context compaction works

Kangwook Lee ingested Apr 27, 2026

Kangwook Lee's prompt-injection probe extracting Codex's hidden compaction & handoff prompts.

你不知道的 Claude Code：架构、治理与工程实践

Tw93 ingested Apr 27, 2026

Tw93's six-layer frame for Claude Code: context, tools, Skills, Hooks, Subagents, verifiers.

你不知道的 Agent：原理、架构与工程实践

Tw93 ingested Apr 27, 2026

Tw93's general-Agent companion essay: Harness > model, ACI tool design, OpenClaw walkthrough.