Harness engineering beginner tutorial, from 0 to 1
TypeScript agentagenticagentic-aiaiai-agentai-agentsdshdsh-pluginharnessharness-engineeringharness-frameworkllm
docs: clarify project 01 experiment criteria
sanbuphya4928cb65ae8
diff --git a/docs/en/projects/project-01-baseline-vs-minimal-harness/index.md b/docs/en/projects/project-01-baseline-vs-minimal-harness/index.mdindex b34fc9f..e5c2b9b 100644--- a/docs/en/projects/project-01-baseline-vs-minimal-harness/index.md+++ b/docs/en/projects/project-01-baseline-vs-minimal-harness/index.md@@ -36,3 +36,10 @@ Repository path: [`projects/project-01/`](https://github.com/walkinglabs/learn-hThe four concrete features are window launch, document list, question panel, andlocal data directory creation. Inspect `solution/feature_list.json` for theexpected evidence for each feature.++This is a comparison experiment, not a requirement that both agent runs+produce a production-ready Electron app. Run the same task against the weak+`starter/` and explicit-harness `solution/`, then record which features each+run completes and what evidence supports the result. Partial or broken output+is valid experimental evidence; the feature list defines what to measure, not+a requirement that the prompt-only run must pass every item.diff --git a/docs/zh/projects/project-01-baseline-vs-minimal-harness/index.md b/docs/zh/projects/project-01-baseline-vs-minimal-harness/index.mdindex a951b49..807c24b 100644--- a/docs/zh/projects/project-01-baseline-vs-minimal-harness/index.md+++ b/docs/zh/projects/project-01-baseline-vs-minimal-harness/index.md@@ -22,6 +22,12 @@四个具体功能是窗口启动、文档列表、问答面板、本地数据目录。预期证据见 `solution/feature_list.json`。+这里是对照实验,不要求两次 agent 运行都产出可用于生产的完整 Electron+应用。请对弱 harness 的 `starter/` 和显式 harness 的 `solution/` 运行同一个+任务,记录各自完成了哪些功能,以及有哪些验证证据。部分完成或运行失败+本身也是实验结果;功能列表用于定义观测指标,并不表示只用提示词运行就必须+全部通过。+## 用什么工具- Claude Code 或 Codex(选一个,两次都用同一个)diff --git a/projects/project-01/README.md b/projects/project-01/README.mdindex f962761..37d8696 100644--- a/projects/project-01/README.md+++ b/projects/project-01/README.md@@ -33,6 +33,17 @@ npm install## Exact Task Contract+Project 01 is a comparison experiment. The primary success criterion is to+run the same agent task against the weak-harness `starter/` and the+explicit-harness `solution/`, then record what each run actually completes.+The generated application does not need to be fully functional for the+comparison to be valid. Use the feature table below to measure completed,+missing, and verified behavior.++The checked-in `solution/` is a reference baseline with the product slice and+harness artifacts already present; it is not a promise that an agent will+recreate every feature from the prompt alone.+The starter prompt is intentionally vague (`starter/task-prompt.md` contains only"Build an Electron app that can show documents and answer questions."). Use thesolution harness to make that vague request concrete: