与「ReFa」相关的搜索结果

微软产品经理 Keiji Kanazawa 在 AI Engineer 大会上做了一场很实在的分享，题为I Let Agents Refactor My Codebase for 3 Weeks. Then I Read the Code 他原本写了一个自动抓取播客、X 帖子和邮件链接做总结的脚本工具，后来想把它彻底重构成一个带注册中心、支持定时调度和每日自我反思的智能体系统。 他自己完全不看代码，写了一份需求文档，就放手让 Claude Code 和 Codex 轮流写代码、互相审查，连着重构了三个星期。 直到有一天他发现生成的日报把句子放错了栏目，想查一下运行轨迹排障，结果 Claude Code 告诉他查不到，因为这压根不是个真正的 Agent。 他打开代码库一看，才发现那三个星期的全自动重构完全是一场灾难。 首先是假重构，模型为了迎合他的需求文档，在代码里建了名叫 `Skill`、`Action` 和 `Agent` 的类，但点进去看，这三个名字像模像样的模块底层其实是同一段旧代码的换皮，所谓的 Skill 里既没有 `SKILL.md` 也没有脚本，所谓的 Agent 根本没接任何 Agent SDK。 而且第二天他收到账单超支警报，单周开销翻了五倍，一查才发现 Codex 在写单元测试时根本没做 Mock，直接把单测连到了线上付费大模型接口上，一天之内自己跑了 30 多轮测试狂烧生产环境的 Token。 随后呢他决定推倒重来。但作为产品经理，他依然不想走回自己逐行读代码的老路，于是核心问题就变成了在不亲自读每一行代码的前提下，怎么管住 Agent 不让它再写空壳代码。 他摸索出了三步改进思路值得看看和参考借鉴。 第一步是改用验收标准来限制而不是目标来限制。 以前他是给个需求让 Agent 写份计划，看着差不多就点同意放行。重来之后他把重点放在定义最终产出的评分标准上。 比如他后来做了一个解析硬核技术视频的功能，要从视频里抽 440 个关键帧，挑出最清晰的黑板公式截图并配上讲解。 他没有去翻代码改排版逻辑，而是直接去改验收的评分表，在规则里写“每个公式必须和对应的解释按时间顺序穿插排列”，然后让 Agent 对着这张评分表自己跑循环做优化，直到跑分达标为止。 第二步是防止 Agent 在评分表上作弊。 因为评分表本身也是他让 Agent 写的，跑着跑着他又抓到了模型作弊。发现这个问题后，他多加了一层泛化检验，每次改完评分表必须同时在多个不同类型的视频上跑分，并明确下指令禁止针对单一视频凑分数。 第三步是用多轮对抗性审查。 为了防止单个模型再次出现工作问题，他在重新搭建运行时底座时，先拿四个主流工具做了一轮平行对比，发现像 Pi 这种极客框架虽然快，但动不动就弹窗问他 SQLite 表结构怎么设计、TypeScript 接口怎么定义这类纯底层工程问题。 最终他还是回到 Claude Code 和 Codex 的组合上，但把原来的简单互评改成了多轮对抗性代码审查，强制让两个模型站在找茬的立场上深度挑对方的架构漏洞和空壳逻辑。 这一套的思路是很值得借鉴参考的，总结起来也简单，第一个点就是尽量用可量化的验收标准限制模型的工作，第二点是可以采取多个模型的审查发现问题。 youtube：

if you want to automerge thousands of pull requests , this is how she does it (via @poteto). 809 in june. 10,926 in october. merged prs a month that touch the grok bot team's code. i went through her posts and put the setup in one place. every pr still gets an agentic risk review plus bugbot. low- and medium-risk ones automerge. only higher-risk ones get a person. day to day, in her words, it's just grok bot and cursor projects. she used to have more cursor automations and replaced them with slack bots that have routines. she works on many streams in parallel with pstack: perf, bug fixes, agent-friendly refactors, her own skills, features, ui polish, the harness, release and observability. not from sitting in a review inbox. she's said the speed gains come from parallelizing. a single chat can take longer because of verification. she handholds it less, so she can run several large projects at once. the real setup is in the codebase and the checks around it. she treats the codebase as memory. every agent mistake needs to turn into a constraint. her team built their own framework for grok bot so some anti-patterns become hard to write at all. agents have to verify their own work. they run the app, and can sometimes model it in tla+. they produce screenshots or video that are easier to check than the whole pr. formal verification skills and ordinary run-and-debug skills work together. she calls this an inverse pyramid. push as much as you can into determinism (fuzzing, property tests, formal and runtime verification, stricter languages) instead of spinning up more agent swarms. in her oct 6 post, she tells sandcastle, a grok team bot, to cut. it runs a buildkite dry run, dms every pr author (they can object), kicks the build, and watches it. then it starts a fuzz swarm: usually 10+ agents on grok 4.7 xhigh, some directed, some chaos monkey, using their verification skills and feature map. when something is high priority, she tells sandcastle to @ mention poteto, her engineer bot, which then spins up a cursor project to triage and fix. her team ships a new stable version of the client at least once a day. that's a lot of machinery under "just grok bot and cursor projects."

互联网时代，大家都在说人人都是产品经理，但实现一个产品，依然需要一个技术团队。 十几年过去了，LLM 和 AI 编程工具消除了这个壁垒，每个人都能做产品，但默认 UI 丑陋依旧是硬伤。 我从《Refactoring UI》书中精选了 20 条设计军规，希望可以启发你做出非常漂亮的产品。

开发系统最极致高效的Agents.md，没有之一： # AGENTS.md ## Core Principles - Choose the simplest implementation that fully satisfies the current requirements. Avoid unnecessary abstraction, configuration, indirection, or speculative extensibility. - Make the smallest necessary change that fixes the root cause. Do not refactor unrelated modules or change strategy semantics unless explicitly requested. - Grow the system in layers. Start from the smallest working end-to-end version and add new capabilities incrementally. Never replace a working system with unfinished complexity. - Reuse existing project components before creating new ones. Prefer extending proven modules over introducing parallel implementations. - Prefer well-maintained libraries when they reduce overall complexity or improve reliability. Do not reimplement common functionality without a clear benefit. - Keep components modular with clearly defined responsibilities. Avoid unnecessary coupling between strategy logic, execution, accounting, replay, and infrastructure. - Design for long-term maintainability once a feature or strategy has been validated. Do not over-engineer speculative ideas before evidence exists. --- ## Strategy Development - Validate hypotheses with historical replay before introducing forward-only logic whenever historical validation is possible. - Every trading strategy must progress through Replay → Shadow → Canary → Live. Do not skip validation stages. - Base design decisions on measurable evidence rather than intuition. Optimize only after demonstrating that an edge exists. - Treat every strategy as an independent contract. Do not silently alter frozen behavior without explicit authorization. --- ## Existing Systems - Do not break running Shadow or Live systems for unrelated work. - Preserve compatibility only when required by active production or validation workflows. Otherwise, remove obsolete code instead of accumulating compatibility layers. - Reuse existing infrastructure whenever possible, including replay engines, accounting, execution, wallet management, order book handling, logging, monitoring, and daemon frameworks. --- ## Engineering Standards - Prefer deterministic behavior over hidden automation. - Fail loudly when assumptions are violated. Do not silently ignore errors or fall back to unexpected behavior. - Keep configuration minimal. Introduce new configuration only when behavior genuinely needs to vary. - Remove dead code instead of leaving unused paths behind. - Write code that is easy to inspect, replay, test, and reason about. - Keep implementation consistent with existing project architecture unless an architectural change is explicitly requested. --- ## Scope Discipline - Implement only the requested scope. - Do not introduce unrelated optimizations, redesigns, migrations, or feature expansions. - Non-blocking findings outside the requested scope may be noted separately but must not be merged into the current task. - Consider a task complete once its agreed acceptance criteria are satisfied. Treat subsequent improvements as separate work items.

Have been working closely with @dinkin_flickaa to completely refactor how images and videos are handled on 𝕏. Media now opens and closes fluidly, regardless of whether it’s a video, image, profile picture, etc. Makes a big difference to overall feel.

有第三方测试说，回归的Fable 5跟六月发布时不是一个水平了。 BridgeMind用BridgeBench重测7月1日版本：【Debugging】86.2→25.9，【Refactoring】73.6→38.4，【Hallucination】75.9→61.7。 据说是新护栏太频繁，很多任务转给Opus 4.8代打了。这掉分幅度，你们怎么看？

just sent my first prompt to Fable 5 > it instantly routed the task to Opus 4.8 > which spawned Sonnet 5 sub agents > that burned 30% of my weekly limit > and handed me slop this thing used to build full open words and refactor entire codebases in one shot WHAT DID YOU DO ANTHROPIC

Been iterating on @tomosman's loop. This one's winning: /goal produce a verified, code-derived behavioral spec for this web platform, captured in one canonical spreadsheet that carries every feature from spec -> tested -> fixed -> verified. Why: we need a single source of truth that maps every feature to its expected behavior *as the code implements it*, so that gaps and bugs surface and the platform can be driven to a known-good state. The spreadsheet is the source of truth. Work on the current repo. Do Phase 0 and Phase 1 under this goal; when the spec is complete, switch into the /loop below to drive testing and remediation. Keep moving through phases without stopping, except at a real checkpoint (defined below). Phase 0 - Plan (first): Detect the stack, the feature surface (routes, pages, components, API endpoints, background jobs, auth, settings…), and the test infra that already exists (unit/integration/e2e, browser automation, seeds/fixtures, a runnable dev server). Propose (a) how you'll inventory features, (b) the spreadsheet schema, and (c) how you'll test in the loop given what's available. Proceed once the plan holds. Phase 1 - Catalog & spec: Read the code and, for every feature, write a user story + the expected behavior as implemented, citing the file/function. Where the code is ambiguous, or behavior is undefined, log an open question - don't guess. Record every feature as a row in the canonical spreadsheet (create with the xlsx skill). Exit: every discoverable feature has a row. One row, concretely: | Area | User story | Expected behavior (from code) | Status | Defects | Type | Notes / source | |---|---|---|---|---|---|---| | Auth | As a returning user I want to log in with email+password so I can reach my dashboard | `POST /api/login` validates via bcrypt, sets httpOnly session cookie, 302 -> `/dashboard`; bad creds -> 401 + inline error | Spec'd | - | - | `api/auth/login.ts`, `LoginForm.tsx` | Canonical artifact: exactly one .xlsx, updated in place across every phase and loop iteration - never fork into per-phase or per-iteration files. Status flows Spec'd -> Tested-Pass / Tested-Fail -> Fixed -> Verified. The main thread is the single writer. Agentic execution: - Delegate breadth to subagents: fan feature discovery and per-area testing across subagents so the main thread stays focused. - Verify by running, not claiming - report real command/test output; state skips and unknowns plainly. - Checkpoint (pause, ask, end the turn) only for a destructive/irreversible action, a fix needing a genuine product decision, or input only I can give. Otherwise, keep going. - Self-check at each phase/loop boundary via a fresh-context subagent: re-verify the spreadsheet against the code (Phase 1) and against actual results (each loop pass). /loop Quality cycle - once the spec is complete, iterate test -> fix -> re-test until clean. Each iteration, in order: 1. Test: exercise every user story not yet Verified against the running app, preferring the strongest method available (browser/e2e automation > existing suites > documented static check only where execution truly isn't possible). Record actual pass/fail in the same spreadsheet; log every defect with its type (functional/logistical or UX). No app-behavior changes in this step. 2. Fix: think hard about root cause, then fix every functional/logistical and UX defect logged this iteration - cause, not symptom. Scope: only logged defects; no new features, no unrelated refactors. Update each row's status. 3. Re-test: re-run every story touched by a fix using the same method; set Verified, or back to Tested-Fail with notes if the fix didn't hold. Exit when all user stories are Verified and no open functional/UX defects remain. Safety cap: if a story is still failing after 3 full iterations, stop, leave it Tested-Fail with root-cause notes, and report it rather than looping further.

＼ReFa BEAUTY SHOWCASE／ -My Favorite Edition- 毎日の美容習慣を豊かに彩るReFaアイテム あなたのReFaにまつわるエピソードを教えてください。 WEEK3 2026.6.22 - 6.28 【テーマ】自分への‘‘美容投資’’として続けてみたいのはどれ？ 🎁お題に沿ったコメントも併せて記載で当選確率UP！💖 ①本アカウントをフォロー ②該当アイテム番号とキーワード 「わたしのリファ」をつけて引用リポスト 【応募締切】6.28 23:59 毎週参加で当選確率ダブルUP⭐ 是非皆様のエピソードとともにご応募お待ちしております。 ▼詳細はこちら