資料來源#
- Anthropic's Boris Cherny: Why Coding Is Solved, and What Comes Next
- Auto mode for Claude Code
- Best Practices for Claude Code
- Full Walkthrough: Workflow for AI Coding — Matt Pocock
- How Anthropic's product team moves faster than anyone else | Cat Wu (Head of Product, Claude Code)
- Introducing Claude Opus 4.7
- The Founder's Playbook: Building an AI-Native Startup
- Zero Trust for AI Agents
摘要#
Anthropic 的 agentic coding 產品，由 Boris Cherny 於 2024 年底在內部 incubator (Anthropic Labs) 中建立。最初是 type-ahead-replacement；隨後以 Sonnet 3.5 pivoted 至 agentic；在 Opus 4（2025 年 5 月）引發拐點前，經歷了約 6 個月的 pre-PMF 階段；隨後在 Opus 4.5、4.6 和 4.7 版本中進一步發展。截至 2026 年 5 月，它已在 CLI、desktop、web、mobile 和 IDE（VS Code、JetBrains）上出貨——相同的產品表面，不同的進入點。
Codebase#
根據 Boris Cherny 的說法（於 2026 年 3 月因 release-process PR 中的人為疏失而公開外洩——隨後已進行強化，非惡意外洩），其實作並無特別之處：TypeScript + React。此選擇是刻意為之：在語言覆蓋率至關重要的時期（2024 年底），這對 model 來說是「非常 on-distribution 的」。
Boris 的聲明：「一年後只需 100 行程式碼」——參見 Harness Shrinkage as Models Improve 以了解其發展方向，而非字面上的聲明。
Surface forms#
- CLI —— 主要表面，最先獲得新功能。「所有 tools 中最強大的一個」（Cat Wu）。
- Desktop app —— 用於 front-end 工作的 preview-pane 整合；對非技術用戶更友善。
- IDE extensions —— VS Code、JetBrains；根據 Boris 的 audience-poll，其使用份額明顯小於 CLI。
- Mobile + web —— 在 AFK 時啟動任務；Boris 的 daily driver。
- Sibling: Cowork —— 相同的 primitives，非程式碼的產出。
Notable features#
/loop—— Claude-scheduled 的 cron job；Agent Loop Pattern 的主要 primitive/powerup—— 涵蓋 100 多個現有功能的 onboarding tour- Sub-agents —— 回報摘要的 token-isolated context windows；參見 Context Window Smart Zone
- Skills —— repo 中可讓 Claude 根據需求拉取的 markdown 檔案；參見 Deep Modules for Agents 中的 push/pull
- Auto mode —— 基於 classifier 的權限限制；參見 Claude Code Auto Mode
- Routines —— 伺服器端的
/loop；即使關閉筆記型電腦仍可繼續運行 - Code review —— slash command + multi-agent 審查模式，根據 Cat Wu 的說法，在 Opus 4.5/4.6 左右變得可靠
- Claude Code Security —— 限制型 beta（2026 年 5 月）；掃描 codebases 以尋找安全漏洞，並提出針對性的 patch 供人工審查。能顯現傳統方法遺漏的 issues。定位於 MVP→Launch 轉型階段（參見 AI-Native Startup Lifecycle）；明確聲明並非人工審查者或合格合規審查的替代品
Security posture (Zero Trust reference implementation)#
Zero Trust for AI Agents 將 Claude Code 列為幾乎每個控制層級的運行範例——其安全 primitives 對應到該框架的八個控制領域：
- Deny-by-default permissions + 限制在專案目錄內的 write-access restrictions → Least Agency / Blast Radius (Agentic)
- Sandboxing（filesystem + network isolation，OS-level enforcement）→ resource boundaries / blast-radius containment
- 用於 MCP 連線的 OAuth 2.0 with auto-refresh、OS credential store、
apiKeyHelper、session-scoped "ask" permissions → Agent Identity and Authentication - Command blocklist（預設為 curl/wget）、用於網頁內容的 isolated context windows、command-injection detection、network-request approval → Agentic Prompt Injection 的防禦
- 預設的 Session isolation +
cleanupPeriodDays+ checkpoint/rewind → Memory and Context Poisoning 安全防護 - Managed settings /
allowManagedPermissionRulesOnly/ 伺服器託管的 (MDM) 設定 → 使用者無法覆寫的組織級治理 - Hooks（PreToolUse 參數驗證、ConfigChange 稽核）+ 帶有單一 session
session.id的 OpenTelemetry 遙測 → 可追溯性與 tool-access control
Notable users (per sources)#
- Boris Cherny —— 100% 的個人程式碼透過 Claude Code 撰寫，單日提交 150 個 PR
- Cat Wu —— 主要的 IC tool，推動功能迭代速度
- Fiona Fung —— 帶領工程與產品團隊；「我們使用 Claude Code 來構建 Claude Code」
- Andrej Karpathy —— power user（"cloud code / codex / open claw"）；將此領域定義為 agentic engineering
- Matt Pocock —— workshop-grade workflow；將 skills 套用於 grill-me + Ralph loop pattern
- Thariq Shihipar —— Claude Code 工程師；「HTML is the new markdown」的 workflows（參見 HTML as the New Markdown）
- Anthropic Applied AI 團隊 —— 僅次於工程團隊的內部第二大 token 消費者
Adjacent / dependent concepts#
- Claude Code Best Practices —— 官方 best-practices 文件
- Claude Code Auto Mode —— 權限 classifier
- Claude Opus 4.7 —— 目前 the model
- Mythos Model —— 內部使用的預覽 model
- Agent Loop Pattern —— loop primitive
- Harness Shrinkage as Models Improve —— Boris 的發展路徑聲明
- Engineer PM Convergence —— 構建它的團隊體現了此趨勢
- Anthropic Labs —— Claude Code 誕生的 incubator（2024 年底）
- Claude Design —— 構建於 Claude Code's Agent SDK + 現有的 Claude Code skill 之上；Claude Design 的 "handoff to Claude Code" 會將設計傳回此處進行實際生產
- AI Native Product Cadence —— Cat Wu 對團隊如何交付產品的闡述
- Agentic Misalignment (AM) —— agent-mode + skip-permissions 的部署暴露了 AM 的威脅表面；與無人值守的
/loop使用者相關 - Claude's Constitution / Model Spec —— 塑造 Claude Code 性格與拒絕行為的價值觀
- AI Employee Framing —— Claude Code 是同一個產品問題的工程工具面，而該問題在 HBR 的 HR/財務情境中被廣泛研究
- HTML as the New Markdown —— 團隊內部的一種 Claude Code 使用模式：將 HTML artifacts 作為面向人類的溝通媒介
- Disposable Micro-Apps / Living Design System —— Thariq 的其他 Claude Code workflows
- Compute Allocator —— Claude Code 越來越多服務的使用者角色：決定什麼值得使用 compute，而不是親自撰寫程式碼
- AI-Native Startup Lifecycle —— 在創辦人四個階段中，作為主要 MVP 構建工具的 Claude Code
- Agentic Technical Debt —— 在沒有持久 CLAUDE.md context 的情況下，使用 Claude Code 特有的失效模式
- Zero-Friction Scope Creep —— 由於 Claude Code 每個功能的低時間成本所導致的失效模式
- Founder as Agent Orchestrator —— Claude Code 為非工程背景創辦人帶來的創辦人角色轉變
- Compounding Data Moat —— Claude Code 構建特定垂直領域的測試套件與整合，作為護城河組件
- Problem-Solution Fit Discipline —— Claude Code 在 Idea 階段的角色被限制在作為討論道具的 lightweight prototype——明確排除驗證用途；此紀律使該工具保持在支援證據的角色中
- Verification as the New Bottleneck —— Fiona Fung：在 Claude Code 團隊中，編寫程式碼已不再是緩慢的部分；驗證/審查才是
- Code as Source of Truth —— Claude Code 團隊的知識共享規範：程式碼 + 簽入的 skills 是 Source of Truth，onboarding 則是透過 Claude 進行
- Vibe Coding vs. Agentic Engineering —— Claude Code 作為其典型表面的領域
- Zero Trust for AI Agents —— 將 Claude Code 列為所有控制層級參考實作的安全框架
- Least Agency / Blast Radius (Agentic) / Agent Identity and Authentication / Agentic Prompt Injection / Memory and Context Poisoning —— Claude Code 的 primitives 所實作的具體 Zero Trust controls（參見上方的安全防護）
資料來源#
- Anthropic's Boris Cherny: Why Coding Is Solved, and What Comes Next
- How Anthropic's product team moves faster than anyone else | Cat Wu (Head of Product, Claude Code)
- Best Practices for Claude Code
- Auto mode for Claude Code
- Introducing Claude Opus 4.7
- Full Walkthrough: Workflow for AI Coding — Matt Pocock
- The Founder's Playbook: Building an AI-Native Startup
Cited by 52
- Agent Identity and Authentication
The foundation control for agentic Zero Trust: cryptographically-rooted per-agent identity (→X.509→hardware attestation…
- Agent Loop Pattern
`/loop` (cron-scheduled) and Ralph Wiggum (backlog-draining) loops as next-generation agent primitive; AFK execution, p…
- Agent-Native Infrastructure
The world is still built for humans and must be rewritten for agents; "what do I copy-paste to my agent?"; sensors/actu…
- Agentic Honesty & Diligence
As models get more capable, failing to surface decision-relevant information shifts from a capability failure to an ali…
- Agentic Misalignment (AM)
Lynch et al. 2025 eval and threat model: LLM email-agent discovers it may be deleted, can take harmful actions; OOD rel…
- Agentic Technical Debt
Debt that *compounds* (not just accumulates) because each agentic-coding session re-derives architectural decisions wit…
- AI Accelerating AI Development
The empirical core of *When AI builds itself*: measured evidence AI already speeds AI R&D at Anthropic — >80% of merged…
- AI Native Product Cadence
Cat Wu's 6mo→1mo→1day cadence at Anthropic: research-preview branding, mission-as-tiebreaker, evergreen launch room, li…
- AI-Native Product Org Bottlenecks
AI-native product-org bottleneck is accountable taste at speed: dogfooding trains taste, evals encode it, and accountab…
- AI-Native Startup Lifecycle
Anthropic's May 2026 reframing of Idea/MVP/Launch/Scale assuming AI infrastructure: each stage's headcount/capital/skil…
- Opinions on Using AI Tools & the Future of the Software Engineering Role
Debate map of four stances on using AI tools (bullish-insider / pragmatist-practitioner / skeptic-governance / architec…
- Andrej Karpathy
Co-founder OpenAI, ex-Tesla AI, Eureka Labs; coined "vibe coding," Software 1/2/3.0, "ghosts not animals," "agentic eng…
- Anthropic
AI safety company / vendor of Claude; mission-as-tiebreaker culture; ~30–40 PMs across teams; Mike Krieger leads Labs r…
- Anthropic Labs
Anthropic's internal incubator — a 'bet factory' of ~a dozen tiny teams exploring the model frontier with lean-startup…
- Boris Cherny
Creator of Claude Code at Anthropic; phone-driven workflow with hundreds of agents; primary advocate of `/loop` primiti…
- Build for the Next Model
Prototype the thing that almost works, not the thing that already works: bet that the next concrete model release (not…
- Building Is Cheap, Arguing Is Expensive
"In technical debate, code wins": generate three PRs vs whiteboard; prototype over design doc; reduce design docs
- Cat Wu
Head of Product for Claude Code and Cowork at Anthropic; primary articulator of AI-native product cadence and engineer-…
- Claude Character as Product
Personality as load-bearing product surface; Amanda's role at Anthropic; lunchtime vibe-checks as eval discipline; the…
- Claude Code Best Practices
Anthropic's guide to effective Claude Code usage: context management, verification-driven development, explore→plan→cod…
- Claude Design
Anthropic Labs product (research preview, ~April 2026) for collaborating with Claude on polished visual artifacts — des…
- Claude Fable 5
Anthropic's first generally-available Mythos-class model (June 2026) — state-of-the-art on nearly all benchmarks; the s…
- Claude Opus 4.7
GA frontier model from Anthropic; direct upgrade to 4.6 at same price; literal instruction following, 1.0–1.35× tokeniz…
- Compounding Data Moat
Anthropic's prescription for Scale-stage defensibility: time-locked behavioral fingerprint + domain-encoded edge cases…
- Compounding Loop Optimization
Dan Carey's discipline of instrumenting and automating every recurring step of the build loop — because when internal t…
- Compute Allocator
The human's evolving role: deciding what's worth spending compute on; ~1% of generated tokens ship, 99% is scaffolding…
- Cowork
Anthropic's non-code knowledge-work agent product; sibling to Claude Code; output is decks/inbox/dossiers; same MCP/com…
- Disposable Micro-Apps
Throwaway custom UIs built per-task to edit a plan ("micro-software on top of micro-software"); copy-back-to-markdown;…
- Where Does Agent Harness Work Remain Durable as Models Improve?
Durable harness work lives at external-reality boundaries: repo-local source of truth, mechanical verification, context…
- Engineer PM Convergence
Generalists across disciplines; product taste as bottleneck skill; Anthropic Claude Code team as case study; "just do t…
- Evals as Product Spec
Cat Wu's framing of evals as the emerging core PM skill: ten great evals beats a hundred mediocre; encode what done loo…
- Fiona Fung
Leads engineering + product for Claude Code and Cowork at Anthropic (ex-Meta/Microsoft); "what served you prior may no…
- Founder as Agent Orchestrator
Founder role shift: less individual contributor, more orchestrator of specialized AI assistants; non-technical founders…
- The Future of Agent Interfaces
Interface future is layered: native interaction models for human collaboration, MCP/APIs for structured action, app pro…
- Harness Shrinkage as Models Improve
Prompt scaffolding shrinks each model release; Cat Wu's pruning discipline; Boris Cherny "100 lines of code a year from…
- HTML as the New Markdown
Thariq Shihipar's thesis: as models improve, thousand-line markdown plans overwhelm the *human*; HTML artifacts (visual…
- Learning to Co-Work with AI: A Software Engineer's Field Guide
Field guide for software engineers in the AI era: 6 skill clusters (taste, harness, alignment-first planning, agent-fri…
- Living Design System
`design_system.html` extracted from repos as a portable, human- and machine-readable source of truth; component playgro…
- Managers as ICs
Every Claude Code manager starts as an IC; flat org; agentic coding collapsed the onboarding cost that pushed managers…
- MCP and Computer Use
Anthropic's two complementary connector mechanisms: MCP for structured programmatic access (Salesforce/Drive/Gmail/Slac…
- Memory and Context Poisoning
Corruption of persistent agent memory that influences behavior long after the initial injection; includes RAG poisoning…
- Entities — People, Orgs, Tools & Projects
Map of Content for all 32 entity pages. See Home for concept domains.
- Open Questions Backlog
_96 pages with open questions, as of 2026-06-14._
- Orchestration vs Employee Framing: Reconciling the Founder's Playbook with HBR's Accountability Evidence
Reconciles the Founder's Playbook orchestration framings with HBR Kropp et al.'s accountability evidence; "orchestratio…
- Problem-Solution Fit Discipline
Idea-stage thesis: three defenses against premature building (time, resources, belief friction) all eroded; AI as devil…
- Recursive Self-Improvement
An AI system autonomously designing and developing its own successor; Anthropic Institute's *When AI builds itself* arg…
- Thariq Shihipar
Engineer on the Claude Code team at Anthropic; "HTML is the new markdown" and "compute allocator" framings; three HTML-…
- Thinking Machines Lab
AI research lab behind interaction models (May 2026); harness-dissolves-into-model thesis; upstreamed streaming-session…
- Verification as the New Bottleneck
Fiona Fung: coding is no longer the bottleneck — verification, review, maintenance are; shift-left; TDD loses its tax;…
- When Does Verification Quality Determine Whether AI Automation Works?
Verification-quality ladder from Lean/formal proof search through software CI and vulnerability reproduction; autonomy…
- Vibe Coding vs. Agentic Engineering
Vibe coding raises the floor (anyone builds); agentic engineering preserves the quality bar while going faster; ">10x a…
- Zero-Friction Scope Creep
MVP failure mode when agentic coding removes the cost-based forcing function against scope creep; antidote is written s…
Related articles
- Harness Shrinkage as Models Improve
Prompt scaffolding shrinks each model release; Cat Wu's pruning discipline; Boris Cherny "100 lines of code a year from…
- Anthropic
AI safety company / vendor of Claude; mission-as-tiebreaker culture; ~30–40 PMs across teams; Mike Krieger leads Labs r…
- Engineer PM Convergence
Generalists across disciplines; product taste as bottleneck skill; Anthropic Claude Code team as case study; "just do t…
- Boris Cherny
Creator of Claude Code at Anthropic; phone-driven workflow with hundreds of agents; primary advocate of `/loop` primiti…
- Cowork
Anthropic's non-code knowledge-work agent product; sibling to Claude Code; output is decks/inbox/dossiers; same MCP/com…