Draft 0.4.1 Consistency & Productization Pass：COMMERCIAL 关系模型（Offering 商业对象、Event/Unit/Quantity 三分离、Consumer Account、Quote/Pricing Policy、Metering Ledger、独立毕业）；Whitepaper 校准（Capability 定义、thesis & assumptions、Emerging Evidence、Value Evidence Ladder、Distribution Events、措辞降确定性）；DATA Verification 解耦 + raw/derived 分层；METRICS M3 grain 修正 + Active Clients 迁移；QUALITY provenance + coverage_basis；全仓 S0-S4/L2/E2/Receipt 术语迁移；product/ 六文档 + Caller Identity Confidence；README 瘦身 + claims 对齐；zh whitepaper 改名

Author: Roy tong

README.md

@@ -13,6 +13,8 @@ AgentMeasure defines a common language for measuring how AI agents discover, cho
 > AgentMeasure is **not** a payment protocol, marketplace, or universal ranking system.
 > It standardizes the facts and measurement semantics those systems can build on.
 
+[Whitepaper](whitepaper/measuring-software-used-by-ai-agents.md) · [白皮书（中文）](whitepaper/measuring-software-used-by-ai-agents.zh-CN.md) · [Core Specification](standard/CORE.md) · [中文](README.zh-CN.md)
+
 ---
 
 ## Why capabilities need a new measurement layer
@@ -65,6 +67,10 @@ Measured Usage ≠ Billable Usage
 
 The last inequality is why the Operation/Attempt model matters commercially: 3 attempts of one operation are not 3 billable operations — unless the metering policy says so.
 
+> The extended thesis — economic units, scarcity, measurement before monetization,
+> metering semantics, and the evidence that commerce is arriving before measurement —
+> lives in [docs/CAPABILITY-ECONOMY.md](docs/CAPABILITY-ECONOMY.md).
+
 ## The measurement view: Reach → Choice → Use → Utility → Value
 
 AgentMeasure defines **metric families**, not a universal KPI. A search capability, a booking API, and a compute job have different value structures.
@@ -123,29 +129,33 @@ Four things a real developer needs to know:
 >
 > AgentMeasure is **not** on the critical request path.
 
+Product architecture (Provider SDK → local buffer → hosted ingestion → dashboard):
+[product/ARCHITECTURE.md](product/ARCHITECTURE.md).
+
 ## What AgentMeasure measures today
 
 - **Decision Opportunity / Candidate Set / Presentation / Selection** — the four objects of choice; the Observed Selection Rate denominator is Presented, not Available
 - **Software Entity → Capability → Interaction Surface** — what exists, what it can do, and the observable interface; observation happens on surfaces, attribution resolves to entities through the machine-readable registry
 - **Operation / Attempt** — one logical use vs. one execution; **retries are multiple attempts of one operation**, not a validity class (they are kept as reliability signals, not counted as distinct logical uses)
 - **Qualified Usage** — production usage after excluding test, benchmark, synthetic, replay, duplicate and other invalid traffic according to policy
-- **Result Consumption / Effect Confirmation** — the result was used by the task, or the intended world-state change was confirmed (Interaction Classes: Information / Action / Transaction / Computation / Communication / …)
+- **Result Consumption** — *defined, reference partial*: the result was used by the task
+- **Effect Confirmation** — *domain model defined, metric planned for Draft 0.5*: the intended world-state change was confirmed
 - **Measurement Label** — a nutrition label for every public number (coverage / sampling / policy / method)
 
 The full model lives in the [Core Specification](standard/CORE.md), [Metrics](standard/METRICS.md), [Entity](standard/ENTITY.md), and [Quality](standard/QUALITY.md). The README is not a spec summary.
 
 ## From measurement to CaaS
 
-```text
-Capability
-    │
-    ▼
-Discover → Choose → Use → Deliver Value → Measure → Meter → Price / Pay / Settle
-```
-
-AgentMeasure standardizes the data and semantics of the **first five steps** — discovery through measurement. Metering and commercial attribution are future extensions; payment rails can be provided by existing payment infrastructure.
+AgentMeasure is **progressively standardizing** the measurement chain from discovery
+and choice through execution, utility and value: core usage semantics are defined
+today; utility/value and commercial metering remain active drafts. Metering and
+commercial attribution are future extensions; payment rails can be provided by
+existing payment infrastructure.
 
 > **AgentMeasure standardizes economic facts, not money movement.**
+>
+> Extended thesis: [docs/CAPABILITY-ECONOMY.md](docs/CAPABILITY-ECONOMY.md) ·
+> Economic semantics: [extensions/COMMERCIAL.md](extensions/COMMERCIAL.md)（Experimental）
 
 ## What AgentMeasure is / is not
 
@@ -173,20 +183,28 @@ AgentMeasure standardizes the data and semantics of the **first five steps** —
 ```bash
 git clone https://github.com/roy-tong/AgentMeasure && cd AgentMeasure
 python3 conformance/runners/run_metrics.py   # metric vectors (M2.2 / M2.5 / M4.1)
-python3 verify_vectors.py                     # receipt / correlation / operation vectors
+python3 verify_vectors.py                     # verification / correlation / operation vectors
 python3 registry/validate_entities.py         # validate the machine-readable registry
 ```
 
 ## Product MVP — in development
 
-The first product path is **Remote MCP / API Capability Measurement**: an AgentMeasure Provider SDK that emits observations from the provider side (no agent-side install), feeding a collector and hosted analytics. The SDK and hosted analytics are not implemented yet — the standard, reference collector, and conformance suite are.
+The first product path is **Remote MCP / API Capability Measurement**: an AgentMeasure
+Provider SDK that emits observations from the provider side (no agent-side install),
+feeding a collector and hosted analytics. The SDK and hosted analytics are not
+implemented yet — the standard, reference collector, and conformance suite are.
+
+Scope and acceptance: [product/MVP.md](product/MVP.md) · SDK contract:
+[product/PROVIDER-SDK.md](product/PROVIDER-SDK.md) · Deployment:
+[product/DEPLOYMENT.md](product/DEPLOYMENT.md)
 
 ## Repository map
 
 ```text
 AgentMeasure/
 ├── standard/          # the normative standard (CORE / METRICS / QUALITY / DATA / ...)
 ├── extensions/        # experimental, non-normative profiles (COMMERCIAL.md)
+├── product/           # product architecture (SDK / hosted analytics, in development)
 ├── whitepaper/        # methodology papers (EN/CN)
 ├── conformance/       # language-neutral test vectors + runners
 ├── reference/         # reference implementation (collector + adapters)

README.zh-CN.md

@@ -13,6 +13,8 @@ AgentMeasure 定义一套共同语言，衡量 AI Agent 如何发现、选择、
 > AgentMeasure **不是**支付协议、市场或通用排名系统。
 > 它标准化的是这些系统可以构建于其上的事实与测量语义。
 
+[Whitepaper](whitepaper/measuring-software-used-by-ai-agents.md) · [中文白皮书](whitepaper/measuring-software-used-by-ai-agents.zh-CN.md) · [Core Specification](standard/CORE.md) · [English](README.md)
+
 ---
 
 ## 为什么 Capability 需要新的测量层
@@ -66,6 +68,9 @@ Agent → Capability → 执行 → 结果
 最后一个不等式正是 Operation/Attempt 模型的商业意义：一次 Operation 的 3 次
 Attempt ≠ 3 次可计费操作——除非计量策略如此规定。
 
+> 完整论点——经济单元、稀缺性、先计量后变现、计量语义，以及"商业先于计量到来"
+> 的证据——见 [docs/CAPABILITY-ECONOMY.md](docs/CAPABILITY-ECONOMY.md)。
+
 ## Measurement View：Reach → Choice → Use → Utility → Value
 
 AgentMeasure 定义 **Metric Families**，不定义全局北极星。搜索能力、预订 API、计算任务的价值结构各不相同。
@@ -128,31 +133,32 @@ Claude / Codex
 >
 > AgentMeasure **不在请求关键路径上**。
 
+产品架构（Provider SDK → 本地缓冲 → 托管 ingestion → 面板）：
+[product/ARCHITECTURE.md](product/ARCHITECTURE.md)。
+
 ## 今天测量什么
 
 - **Decision Opportunity / Candidate Set / Presentation / Selection** —— 选择的四个对象；Observed Selection Rate 的分母是 Presented，不是 Available
 - **Software Entity → Capability → Interaction Surface** —— 存在什么、能做什么、怎么交互；观察发生在 surface 层，归属经机器可读 registry 解析到 entity
 - **Operation / Attempt** —— 一次逻辑使用 vs 一次执行；**重试 = 同一 Operation 的多个 Attempt，不再是 validity 分类**（作为可靠性信号保留，不当作多次逻辑使用计数）
 - **Qualified Usage** —— 按策略排除 test / benchmark / synthetic / replay / duplicate 等无效流量后的生产使用
-- **Result Consumption / Effect Confirmation** —— 结果被任务使用，或预期的世界状态改变被确认（Interaction Classes：Information / Action / Transaction / Computation / Communication / …）
+- **Result Consumption** —— *已定义，参考实现部分*：结果被任务使用
+- **Effect Confirmation** —— *领域模型已定义，指标计划于 Draft 0.5*：预期的世界状态改变被确认
 - **Measurement Label** —— 每个公开数字的营养成分表（覆盖 / 采样 / 口径 / 方法）
 
 完整模型见 [Core Specification](standard/CORE.md) / [Metrics](standard/METRICS.md) /
 [Entity](standard/ENTITY.md) / [Quality](standard/QUALITY.md)。README 不是 spec 摘要。
 
 ## 从 Measurement 到 CaaS
 
-```text
-Capability
-    │
-    ▼
-Discover → Choose → Use → Deliver Value → Measure → Meter → Price / Pay / Settle
-```
-
-AgentMeasure 标准化**前五步**——从发现到测量——产生的数据与语义。计量与商业归因
-是未来扩展；支付轨道可以由现有支付基础设施提供。
+AgentMeasure 正在**渐进标准化**从发现与选择、经执行到效用与价值的测量链：核心使用
+语义今天已定义；效用/价值与商业计量仍是活跃草案。计量与商业归因是未来扩展；支付
+轨道可以由现有支付基础设施提供。
 
 > **AgentMeasure 标准化经济事实，不移动金钱。**
+>
+> 完整论点：[docs/CAPABILITY-ECONOMY.md](docs/CAPABILITY-ECONOMY.md) ·
+> 经济语义：[extensions/COMMERCIAL.md](extensions/COMMERCIAL.md)（Experimental）
 
 ## AgentMeasure 是 / 不是
 
@@ -180,7 +186,7 @@ AgentMeasure 标准化**前五步**——从发现到测量——产生的数据
 ```bash
 git clone https://github.com/roy-tong/AgentMeasure && cd AgentMeasure
 python3 conformance/runners/run_metrics.py   # 指标 vectors（M2.2 / M2.5 / M4.1）
-python3 verify_vectors.py                     # receipt / correlation / operation vectors
+python3 verify_vectors.py                     # verification / correlation / operation vectors
 python3 registry/validate_entities.py         # 校验机器可读 registry
 ```
 
@@ -190,12 +196,17 @@ python3 registry/validate_entities.py         # 校验机器可读 registry
 SDK 在 Provider 侧产出 observations（无需 Agent 侧安装），接入 collector 与托管分析。
 SDK 与托管分析尚未实现——标准、参考 collector 与 conformance 套件已就绪。
 
+范围与验收：[product/MVP.md](product/MVP.md) · SDK 契约：
+[product/PROVIDER-SDK.md](product/PROVIDER-SDK.md) · 部署：
+[product/DEPLOYMENT.md](product/DEPLOYMENT.md)
+
 ## 仓库结构
 
 ```text
 AgentMeasure/
 ├── standard/          # 规范性标准本体（CORE / METRICS / QUALITY / DATA / ...）
 ├── extensions/        # 实验性、非规范性 profile（COMMERCIAL.md）
+├── product/           # 产品架构（SDK / 托管分析，in development）
 ├── whitepaper/        # 方法论论文（中英）
 ├── conformance/       # 语言无关 test vectors + runners
 ├── reference/         # 参考实现（collector + adapters）

docs/CAPABILITY-ECONOMY.md

@@ -0,0 +1,86 @@
+# The Agent Capability Economy — Extended Thesis
+
+> 这是项目 Vision 的展开叙事（非规范性）。README 只保留摘要；
+> 规范定义见 `standard/`，经济语义见 `extensions/COMMERCIAL.md`（Experimental）。
+
+## 1. 经济单位在变（增量，不是替代）
+
+```text
+SaaS
+Human → Application → Seat / Month
+
+API Economy
+Software → API → Request / Token
+
+Agent Capability Economy
+Agent → Capability → Operation / Outcome
+```
+
+**Alongside** seat-based SaaS and request-based APIs, callable capabilities are
+emerging as a new economic unit for agent-mediated software consumption. Seats will
+not disappear; requests will not disappear; capability/outcome units add a new layer.
+
+## 2. 稀缺性下移
+
+接口（Skill / MCP adapter / CLI）越来越廉价易造；稀缺层是接口背后的访问权：
+
+```text
+Data · Compute · Action · Permission · Trust · Real-world fulfillment
+```
+
+> **Interfaces may become cheap to create; capabilities remain scarce to deliver.**
+
+## 3. 计量先于变现
+
+CaaS 要能定价、计费与建立声誉，先要有共同的测量语义：
+
+```text
+1 user task → 1 Operation → 3 retries：收 1 次还是 3 次？（Metering Policy 决定）
+返回成功但 Agent 忽略结果：价值交付了吗？（Consumption / Effect）
+预订 API 执行了但从未确认：能力履约了吗？（Effect Confirmation）
+任务成功了但没有这个能力也能成功：Provider 能主张价值吗？（Incrementality）
+```
+
+**Measurement before monetization** —— 这就是 AgentMeasure 的 wedge。
+
+## 4. 现实证据：商业先于计量到来
+
+- Cloudflare Agents SDK：MCP Tool 按调用定价并经 x402 收费
+  （[Charge for MCP tools](https://developers.cloudflare.com/agents/agentic-payments/x402/charge-for-mcp-tools/)）
+- Coinbase x402 Bazaar：Agent 搜索带价格/schema 的服务并经 MCP 付费调用
+  （[x402 Bazaar](https://docs.cdp.coinbase.com/x402/bazaar)）
+- OpenAI × Stripe Agentic Commerce Protocol（ACP）：真实 agentic commerce 流程
+  （[报道](https://www.digitaltransactions.net/openai-and-stripe-are-the-latest-fintechs-to-enable-agentic-commerce/)）
+
+> **Payment and discovery infrastructure is arriving before common capability
+> measurement semantics.**
+
+## 5. 从测量到计费（语义链）
+
+```text
+Capability → Discover → Choose → Use → Deliver Value → Measure → Meter → Price / Pay / Settle
+```
+
+- AgentMeasure 渐进标准化前五步产生的数据与语义（当前已定义：Core usage semantics）
+- Metering / commercial attribution 是未来扩展（extensions/COMMERCIAL.md）
+- Payment rails 由现有支付基础设施提供
+
+> **AgentMeasure standardizes economic facts, not money movement.**
+
+## 6. 三条线永不耦合
+
+| 线 | 负责 | 文档 |
+| --- | --- | --- |
+| Core | 测事实 | standard/（规范性） |
+| Commercial Extension | 事实 → 经济单位的解释 | extensions/COMMERCIAL.md（实验性） |
+| Product | 帮用户获得这些事实 | product/（in development） |
+| Payment 系统 | 钱 | 外部 |
+
+**CaaS 是 Vision，不是 Core Standard 成立的必要条件**：AgentMeasure Core 1.0
+独立毕业为 Agent 软件测量标准。
+
+## 7. 叙事锁定
+
+- **Today**：让开发者知道 Agent 如何真实使用自己的能力（Remote Capability Analytics）
+- **Direction**：让 Capability 可以跨 Agent 被统一度量、比较和计量
+- **Vision**：成为 CaaS 与 Agent Capability Economy 的统一计量基础

docs/adapters.md

@@ -2,23 +2,25 @@
 
 > 每个 adapter 能可靠提供什么、不能提供什么——公开声明局限是测量项目的可信度基础。
 > ✅ = 稳定可靠 · ⚠️ = 部分/受限 · ❌ = 不提供 · ? = 待验证
+> 词汇（Draft 0.4.1）：Selection / Operation / Attempt / Completion / Consumption；
+> 生命周期阶段 = selected · invoked · completed · consumed。
 
 | Capability | Codex Hook | Claude OTel | DSH Plugin | MCP Server (wrapper) |
 | --- | --- | --- | --- | --- |
-| Selected（L0） | ✅ | ✅ | ✅ | — |
-| Executed（L1） | ✅ | ✅ | ✅ | ✅ |
-| Success（L2 outcome） | ⚠️ 不可靠（Bash 非零退出仍触发 PostToolUse） | ✅ | ✅ | ✅ |
+| Selection（生命周期 selected） | ✅ | ✅ | ✅ | — |
+| Executed（invoked） | ✅ | ✅ | ✅ | ✅ |
+| Completion outcome（completed） | ⚠️ 不可靠（Bash 非零退出仍触发 PostToolUse） | ✅ | ✅ | ✅ |
 | Duration | ⚠️ 无官方时间戳 | ✅ | ✅ | ✅ |
 | Tool call ID | ✅ `tool_use_id` | ✅ `gen_ai.tool.call.id` | ✅ `callId` | depends on server |
 | Trace ID | ❌ 官方 schema 无 | ✅ | depends | ✅ |
-| Result Consumed（S3） | ❌ | ✅ **MCP 消费信号**（`mcp_server.name`/`mcp_tool.name` 出现在后续 request telemetry） | ? | — |
+| Consumption（result consumed） | ❌ | ✅ **MCP 消费信号**（`mcp_server.name`/`mcp_tool.name` 出现在后续 request telemetry） | ? | — |
 | Agent-side observation | ✅ | ✅ | ✅ | ❌ |
 | Server-side observation | ❌ | ❌ | ❌ | ✅ |
-| 独立构成 E2 | ❌（单侧） | ❌（单侧） | ❌（单侧，harness 内配对仅生命周期） | ❌（单侧） |
+| 独立构成 cross-side corroborated | ❌（单侧） | ❌（单侧） | ❌（单侧，harness 内配对仅生命周期） | ❌（单侧） |
 
 ## 说明
 
 - **Codex Hook 能力边界**来自官方 `PostToolUse` schema：`session_id / model / turn_id / tool_name / tool_use_id`；无 trace_id、无可靠时间戳、无成败判定。
-- **Claude Code 是 S3（Result Consumed）的第一个实证平台**：`claude_code.tool_result` span 提供 `tool_name/tool_use_id/success/duration_ms/error_type`；v2.1.222+ 的 API request telemetry 在**实际消费了某个 MCP tool result** 时才带 `mcp_server.name/mcp_tool.name`——tool_result（tool_use_id=X）→ 下一次 request（mcp_tool.name）即消费证据链。
-- **DSH Plugin**：`tool/call ↔ tool/result` 的 harness 内配对只证明生命周期完成（L2），不构成独立佐证；evidence 由 verifier 计算。
-- **E2 需要跨侧**：任何单侧 adapter 都无法独立构成 E2；E2 = 同一 invocation 的 ≥2 条独立 observer 观察。
+- **Claude Code 是 Consumption 的第一个实证平台**：`claude_code.tool_result` span 提供 `tool_name/tool_use_id/success/duration_ms/error_type`；v2.1.222+ 的 API request telemetry 在**实际消费了某个 MCP tool result** 时才带 `mcp_server.name/mcp_tool.name`——tool_result（tool_use_id=X）→ 下一次 request（mcp_tool.name）即消费证据链。
+- **DSH Plugin**：`tool/call ↔ tool/result` 的 harness 内配对只证明生命周期完成（completed），不构成独立佐证；evidence 由 verifier 计算。
+- **cross-side corroborated 需要跨侧**：任何单侧 adapter 都无法独立构成；同一 attempt 的 ≥2 条独立 observer 观察。