docs: update READMEs with protocol positioning, align EN/ZH structure

- Add File Injection row (Claude Code Memory) to pattern comparison table
- Add protocol gap paragraph (MCP/ODBC/JDBC analogy, three primitives)
- Add intent-native protocol bullet to Features
- Add memory gateway paragraph to Vision
- Add Graph-LLM structural insight to References
- Align ZH with EN: add Go Report Card badge, Claude Max/Pro callout,
  move "from source" under Install, update OpenClaw to --yes with link,
  add NanoClaw skill location

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Author: Grivn

README.md

@@ -28,9 +28,12 @@ Most memory tools embed their own LLM inside the pipeline. Mnemon takes a differ
 | Pattern | LLM Role | Representative |
 |---|---|---|
 | **LLM-Embedded** | Executor inside the pipeline | Mem0, Letta |
+| **File Injection** | None — reads file at session start | Claude Code Memory |
 | **MCP Server** | Tool provider via MCP protocol | claude-mem |
 | **LLM-Supervised** | External supervisor of a standalone binary | **Mnemon** |
 
+Mnemon also addresses a gap in the protocol stack. MCP standardizes how LLMs discover and invoke tools. ODBC/JDBC standardizes how applications access databases. But how LLMs interact with databases using memory semantics — this layer has no protocol. Mnemon's three primitives — `remember`, `link`, `recall` — form an intent-native protocol: command names map to the LLM's cognitive vocabulary (`remember` not INSERT, `recall` not SELECT), and output is structured JSON with signal transparency rather than raw database rows.
+
 <p align="center">
   <img src="docs/diagrams/llm-supervised-concept.jpg" width="720" alt="LLM-Supervised Architecture — three patterns compared, with detailed Mnemon implementation showing hooks, brain/organ split, and sub-agent delegation" />
   <br />
@@ -145,6 +148,7 @@ You don't run mnemon commands yourself. The agent does — driven by hooks and g
 - **LLM-supervised** — the host LLM decides what to remember, update, and forget; no embedded LLM, no API keys
 - **Hook-based integration** — four lifecycle hooks: Prime (load guide), Remind (recall & remember), Nudge (remember), and Compact (save before compression)
 - **Four-graph architecture** — temporal, entity, causal, and semantic edges, not just vector similarity
+- **Intent-native protocol** — three primitives (`remember`, `link`, `recall`) map to the LLM's cognitive vocabulary, not database syntax; structured JSON output with signal transparency
 - **Intent-aware recall** — graph traversal + optional vector search (RRF fusion), enabled by default for all queries
 - **Built-in deduplication** — `remember` auto-detects duplicates and conflicts; skips or auto-replaces
 - **Retention lifecycle** — importance decay, access-count boosting, and garbage collection
@@ -168,6 +172,8 @@ All your local agentic AIs — across sessions and frameworks — sharing one po
 
 The foundation is in place: a single `~/.mnemon` database that any agent can read and write. Claude Code's hook integration is the reference implementation; OpenClaw uses a plugin-based approach; NanoClaw integrates via container skills and volume mounts. The same pattern can be replicated for any LLM CLI that supports event hooks or system prompts.
 
+The longer-term direction is a **memory gateway**: protocol decoupled from storage engine. The current SQLite backend is the first adapter; the protocol surface (`remember / link / recall`) can sit on top of PostgreSQL, Neo4j, or any graph database. Agent-side optimization (when to recall, what to remember) and storage-side optimization (indexing, graph algorithms) evolve independently. See [Future Direction](docs/design/08-decisions.md#82-future-direction) for details.
+
 ## FAQ
 
 **Do different sessions share memory?**
@@ -228,10 +234,11 @@ make help           # show all targets
 
 ## References
 
-Mnemon combines the paradigm of one paper with the methodology of another. See [Theoretical Foundations](docs/DESIGN.md#24-theoretical-foundations) for details.
+Mnemon combines the paradigm of one paper with the methodology of another, grounded in the structural insight that graph memory is isomorphic to LLM attention. See [Theoretical Foundations](docs/DESIGN.md#25-theoretical-foundations) for details.
 
 - **RLM** — Zhang, Kraska & Khattab. [Recursive Language Models](https://arxiv.org/abs/2512.24601). 2025. Establishes the paradigm: LLMs are more effective as orchestrators of external environments than as direct data processors.
 - **MAGMA** — Zou et al. [A Multi-Graph based Agentic Memory Architecture](https://arxiv.org/abs/2601.03236). 2025. Provides the methodology: four-graph model (temporal, entity, causal, semantic) with intent-adaptive retrieval.
+- **Graph-LLM Structural Insight** — Joshi & Zhu. [Building Powerful GNNs from Transformers](https://arxiv.org/abs/2506.22084). 2025; and the Graph-based Agent Memory survey (Chang Yang et al., 2026). Confirms that LLM attention is computationally equivalent to GNN operations — graph memory is a structural match, not an engineering convenience.
 
 ## License
 

docs/zh/README.md

@@ -10,6 +10,7 @@
 
 [![Go 1.24+](https://img.shields.io/badge/Go-1.24%2B-00ADD8?logo=go&logoColor=white)](https://go.dev/)
 [![CI](https://github.com/mnemon-dev/mnemon/actions/workflows/ci.yml/badge.svg)](https://github.com/mnemon-dev/mnemon/actions/workflows/ci.yml)
+[![Go Report Card](https://goreportcard.com/badge/github.com/mnemon-dev/mnemon)](https://goreportcard.com/report/github.com/mnemon-dev/mnemon)
 [![License: MIT](https://img.shields.io/badge/License-MIT-green.svg)](../../LICENSE)
 
 ---
@@ -18,16 +19,21 @@ LLM 智能体在会话之间会遗忘一切。上下文压缩丢失关键决策
 
 Mnemon 为你的 LLM 提供持久的跨会话记忆 — 四图知识存储、意图感知检索、重要度衰减、自动去重。单一二进制，零 API 密钥，一条命令完成部署。
 
+> **Claude Max / Pro 订阅用户？** Mnemon 完全通过你现有的订阅运作——不需要额外的 API 密钥。你的 LLM 订阅*本身*就是智能层。两条命令即可完成。
+
 ### 为什么选择 Mnemon？
 
 多数记忆工具在管线内嵌入自己的 LLM。Mnemon 采用不同路线：**你的宿主 LLM 就是监督者。** 二进制处理确定性计算（存储、图索引、搜索、衰减）；LLM 做判断（记什么、怎么关联、何时遗忘）。没有中间人，没有额外推理开销。
 
 | 模式 | LLM 角色 | 代表项目 |
 |---|---|---|
 | **LLM-Embedded** | 管线内部的执行者 | Mem0, Letta |
+| **File Injection** | 无 — 会话启动时读取文件 | Claude Code Memory |
 | **MCP Server** | 通过 MCP 协议提供工具 | claude-mem |
 | **LLM-Supervised** | 独立二进制的外部监督者 | **Mnemon** |
 
+Mnemon 同时填补了协议栈中的空白。MCP 标准化了 LLM 如何发现和调用工具，ODBC/JDBC 标准化了应用如何访问数据库，但 LLM 以记忆语义与数据库交互——这一层尚无协议。Mnemon 的三个原语——`remember`、`link`、`recall`——构成一个意图原生协议：命令名称映射到 LLM 的认知词汇（`remember` 而非 INSERT，`recall` 而非 SELECT），输出是带有信号透明度的结构化 JSON，而非原始数据库行。
+
 <p align="center">
   <img src="../diagrams/llm-supervised-concept.jpg" width="720" alt="LLM 监督式架构 — 三种模式对比，及 Mnemon 实现细节：钩子、大脑/器官分离、Sub-agent 委派" />
   <br />
@@ -60,6 +66,13 @@ brew install mnemon-dev/tap/mnemon
 go install github.com/mnemon-dev/mnemon@latest
 ```
 
+**从源码构建**：
+
+```bash
+git clone https://github.com/mnemon-dev/mnemon.git && cd mnemon
+make install
+```
+
 **验证安装**：
 
 ```bash
@@ -74,16 +87,13 @@ mnemon setup
 
 `mnemon setup` 自动检测 Claude Code，交互式部署技能文件、钩子和行为引导。启动新会话 — 记忆自动运作。
 
-### OpenClaw
+### [OpenClaw](https://github.com/openclaw/openclaw)
 
 ```bash
-go install github.com/mnemon-dev/mnemon@latest
-mnemon setup --target openclaw
+mnemon setup --target openclaw --yes
 ```
 
-部署技能文件和行为引导到 `~/.mnemon/prompt/guide.md`。由于 OpenClaw 的钩子集成暂未自动化，需要手动配置：
-
-> 阅读 `~/.mnemon/prompt/guide.md` 并按照其 recall/remember 工作流配置自身。
+一条命令将技能文件、钩子、插件和行为引导部署到 `~/.openclaw/`。重启 OpenClaw 网关即可激活。
 
 ### [NanoClaw](https://github.com/qwibitai/nanoclaw)
 
@@ -93,12 +103,7 @@ NanoClaw 在 Linux 容器内运行智能体。使用 `/add-mnemon` 技能集成
 2. 在 NanoClaw 项目中运行 `/add-mnemon` — Claude Code 将修改 Dockerfile、添加容器技能、配置卷挂载
 3. 每个 WhatsApp 群组获得独立的记忆存储，可选全局共享记忆（只读）
 
-### 从源码构建
-
-```bash
-git clone https://github.com/mnemon-dev/mnemon.git && cd mnemon
-make install && mnemon setup
-```
+技能文件位于 NanoClaw 仓库的 `.claude/skills/add-mnemon/` 目录。
 
 ### 卸载
 
@@ -143,6 +148,7 @@ mnemon setup --eject
 - **LLM 监督式** — 宿主 LLM 主动决定记什么、更新什么、遗忘什么；无内嵌 LLM，无 API 密钥
 - **钩子集成** — 四个生命周期钩子：Prime（加载引导）、Remind（recall 和 remember）、Nudge（remember）、Compact（压缩前保存）
 - **四图架构** — 时序、实体、因果、语义四种边，不仅仅是向量相似度
+- **意图原生协议** — 三个原语（`remember`、`link`、`recall`）映射到 LLM 的认知词汇而非数据库语法；结构化 JSON 输出，带信号透明度
 - **意图感知召回** — 图遍历 + 可选向量搜索（RRF 融合），所有查询默认启用
 - **内置去重** — `remember` 自动检测重复和冲突；跳过或自动替换
 - **保留度生命周期** — 重要性衰减、访问计数提升、免疫规则、垃圾回收
@@ -166,6 +172,8 @@ mnemon setup --eject
 
 基础已就绪：一个 `~/.mnemon` 数据库，任何 agent 都可以读写。Claude Code 的钩子集成是参考实现；OpenClaw 使用插件方式集成；NanoClaw 通过容器技能和卷挂载集成。同样的模式可以复制到任何支持事件钩子或系统提示的 LLM CLI。
 
+更长远的方向是**记忆网关**：协议层与存储引擎解耦。当前 SQLite 后端是第一个适配器；协议面（`remember / link / recall`）可运行在 PostgreSQL、Neo4j 或任何图数据库之上。Agent 侧优化（何时召回、记什么）与存储侧优化（索引、图算法）独立演进。详见[未来方向](design/08-decisions.md#82-未来方向)。
+
 ## 常见问题
 
 **不同会话共享记忆吗？**
@@ -225,10 +233,11 @@ make help           # 显示所有目标
 
 ## 参考文献
 
-Mnemon 取用了一篇论文的范式和另一篇论文的方法论。详见[理论基础](DESIGN.md#24-理论基础)。
+Mnemon 取用了一篇论文的范式和另一篇论文的方法论，并基于图记忆与 LLM 注意力同构这一结构洞察。详见[理论基础](DESIGN.md#25-理论基础)。
 
 - **RLM** — Zhang, Kraska & Khattab. [Recursive Language Models](https://arxiv.org/abs/2512.24601). 2025. 建立范式：LLM 作为外部环境的 orchestrator 比直接处理数据更有效。
 - **MAGMA** — Zou et al. [A Multi-Graph based Agentic Memory Architecture](https://arxiv.org/abs/2601.03236). 2025. 提供方法论：四图模型（temporal、entity、causal、semantic）+ intent-adaptive retrieval。
+- **Graph-LLM 结构洞察** — Joshi & Zhu. [Building Powerful GNNs from Transformers](https://arxiv.org/abs/2506.22084). 2025；及图智能体记忆综述（Chang Yang et al., 2026）。证实 LLM 注意力机制在计算上等价于 GNN 操作——图记忆是结构性匹配，而非工程便利。
 
 ## 许可证