EvoClaw — 系统架构图

flowchart LR classDef frontend fill:#2a1f4a,stroke:#a78bfa,color:#fff classDef gateway fill:#1a2f4a,stroke:#5fb3ff,color:#fff classDef loop fill:#0f3a44,stroke:#22d3ee,color:#fff classDef cap fill:#0f3a2a,stroke:#34d399,color:#fff classDef route fill:#3a2f0a,stroke:#facc15,color:#fff classDef policy fill:#3a210a,stroke:#fb923c,color:#fff classDef store fill:#3a1414,stroke:#f87171,color:#fff classDef ext fill:#1a2f4a,stroke:#5fb3ff,color:#fff,stroke-dasharray:4 3 CLI[CLI evoclaw]:::frontend REPL[交互式 REPL]:::frontend GW[本地 Gateway HTTP]:::gateway LOOP[Agent Loop]:::loop RED[Redactor 保险柜]:::policy TOOL[Tool Registry]:::cap SK[Skill Tree]:::cap MEM[记忆 L0 至 L5]:::cap REF[反思 + 蒸馏]:::cap PROV[Provider 适配器]:::route ACP[ACP Provider]:::route MCP[MCP Tool 包装器]:::route POL[权限引擎]:::policy COST[Cost 引擎]:::policy AUD[JSONL Audit]:::policy FS[(workspace 与 logs)]:::store YAML[(skills 下的 yaml)]:::store VAULT[(secrets 下的 vault.json)]:::store AGENTS[(agents 下的 toml)]:::store MCPCFG[(mcp 下的 toml)]:::store MODEL((模型 API)):::ext EXTAGENT((外部 CLI)):::ext EXTSRV((MCP 服务器)):::ext CLI --> REPL REPL --> RED GW --> RED RED --> LOOP LOOP --> TOOL LOOP --> SK LOOP --> MEM LOOP --> REF TOOL --> PROV TOOL --> MCP LOOP --> ACP PROV --> MODEL ACP -. spawn stdio .-> EXTAGENT MCP -. spawn stdio .-> EXTSRV TOOL --> POL PROV --> COST LOOP --> AUD TOOL --> AUD POL --> AUD REF --> SK REF --> MEM SK --> YAML MEM --> FS AUD --> FS RED --> VAULT ACP --> AGENTS MCP --> MCPCFG

flowchart LR U[用户输入] --> RED1[Redactor scrub] RED1 --> TASK[写 Task 记录] TASK --> PLAN[规划这一轮] PLAN --> CACHE{schema 指纹命中?} CACHE -- 是 --> SKIP[复用短标记] CACHE -- 否 --> FULL[发完整 schema] SKIP --> CALL[provider.stream] FULL --> CALL CALL --> ASSIST[assistant 文本 + tool calls] ASSIST --> RED2[Redactor scrub] RED2 --> TOOLS{有工具调用?} TOOLS -- 是 --> EXEC[执行工具] EXEC --> RED3[Redactor scrub 结果] RED3 --> TURN[写 Turn 记录] TURN --> PLAN TOOLS -- 否 --> FINAL[最终文本] FINAL --> REF[反思回合] REF --> DIST[蒸馏] DIST --> SK[Skill upsert] REF --> MEM[记忆 L3 写入] REF --> END[写 End 记录] END --> DOC[doctor closure 审计]

flowchart LR classDef vault fill:#3a1414,stroke:#f87171,color:#fff classDef pat fill:#3a2f0a,stroke:#facc15,color:#fff classDef pass fill:#0f3a2a,stroke:#34d399,color:#fff IN[入口文本] --> V{Vault 命中?} V -- 是 --> SUB["dollar SECRET NAME 占位符"]:::vault V -- 否 --> P{模式命中?} P -- 是 --> RED["REDACTED kind fp"]:::pat P -- 否 --> THRU[原样通过]:::pass SUB --> OUT[模型 / JSONL / 记忆层] RED --> OUT THRU --> OUT

flowchart LR T1[招式 1 schema 哈希指纹] --> G1[省 25 至 40 percent prompt] T2[招式 2 ephemeral cache] --> G2[实际花费降 60 至 70 percent] T3[招式 3 头加尾截断] --> G3[observation 省 50 至 70 percent] T4[招式 4 summary 工作记忆] --> G4[history 省 80 至 90 percent] T5[招式 5 tag 周期压缩] --> G5[长任务省 50 至 60 percent] G1 --> SUM[长任务花费 1/3 以下] G2 --> SUM G3 --> SUM G4 --> SUM G5 --> SUM

flowchart TB subgraph LM["你的笔记本"] direction LR UI[CLI / REPL] GD[evo-gateway daemon] RT[ConversationRuntime] BR[ACP 子进程] MS[MCP 子进程] DATA[(~/.evoclaw)] UI --> RT GD --> RT RT -. spawn .-> BR RT -. spawn .-> MS RT --> DATA end EXT((模型 API)) RT --> EXT