配置参考
RsClaw 使用 JSON5 格式，支持注释、尾逗号和无引号键名。
配置文件位置
| 优先级 | 来源 |
|---|---|
| 1（最高） | --config-path <file> CLI 参数 |
| 2 | $RSCLAW_BASE_DIR/rsclaw.json5 |
| 3 | ~/.rsclaw/rsclaw.json5 |
| 4（最低） | .rsclaw.json5（当前目录） |
配置示例
以下是完整的默认配置模板（源文件：examples/default.rsclaw.json5）。所有段落都是可选的，按需取消注释即可。
json5
// rsclaw.json5 — Default configuration template
// All sections are optional. Uncomment and modify as needed.
// JSON5 syntax: comments, trailing commas, unquoted keys.
// Environment variables: ${VAR} syntax supported in string fields.
// Secrets: use { source: "env", id: "VAR" } for sensitive values.
{
// =================================================================
// GATEWAY — HTTP server, networking, authentication
// =================================================================
gateway: {
port: 18888, // TCP port (default: 18888)
bind: "loopback", // "loopback", "all", "custom", or IP address
// bindAddress: "192.168.1.100", // Custom IP (when bind is "custom" or IP)
language: "Chinese", // Default: "English". Options: Chinese, Japanese, Korean, etc.
processingTimeout: 60, // Seconds before "Processing..." indicator (0=disabled)
// auth: {
// mode: "token",
// token: "${RSCLAW_AUTH_TOKEN}", // Or: { source: "env", id: "RSCLAW_AUTH_TOKEN" }
// },
},
// =================================================================
// MODELS — LLM provider registry
// =================================================================
models: {
providers: {
// anthropic: {
// apiKey: "${ANTHROPIC_API_KEY}",
// },
// openai: {
// apiKey: "${OPENAI_API_KEY}",
// },
// ollama: {
// api: "ollama",
// baseUrl: "http://localhost:11434/v1",
// apiKey: "ollama",
// },
// === 国产大模型配置示例 ===
// DeepSeek (https://platform.deepseek.com)
// deepseek: {
// baseUrl: "https://api.deepseek.com/v1",
// apiKey: "${DEEPSEEK_API_KEY}",
// models: [
// { id: "deepseek-chat", maxTokens: 8192, contextWindow: 64000 },
// { id: "deepseek-reasoner", maxTokens: 8192, contextWindow: 64000 },
// ],
// },
// 阿里云通义千问 (https://dashscope.console.aliyun.com)
// qwen: {
// baseUrl: "https://dashscope.aliyuncs.com/compatible-mode/v1",
// apiKey: "${DASHSCOPE_API_KEY}",
// models: [
// { id: "qwen-turbo", maxTokens: 6144, contextWindow: 131072 },
// { id: "qwen-plus", maxTokens: 6144, contextWindow: 131072 },
// { id: "qwen-max", maxTokens: 6144, contextWindow: 32768 },
// { id: "qwen-long", maxTokens: 6144, contextWindow: 1048576 },
// ],
// },
// 月之暗面 Kimi (https://platform.moonshot.cn)
// kimi: {
// baseUrl: "https://api.moonshot.cn/v1",
// apiKey: "${MOONSHOT_API_KEY}",
// models: [
// { id: "moonshot-v1-8k", maxTokens: 4096, contextWindow: 8192 },
// { id: "moonshot-v1-32k", maxTokens: 4096, contextWindow: 32768 },
// { id: "moonshot-v1-128k", maxTokens: 4096, contextWindow: 131072 },
// ],
// },
// 智谱 GLM (https://open.bigmodel.cn)
// zhipu: {
// baseUrl: "https://open.bigmodel.cn/api/paas/v4",
// apiKey: "${ZHIPU_API_KEY}",
// models: [
// { id: "glm-4", maxTokens: 4096, contextWindow: 131072 },
// { id: "glm-4-flash", maxTokens: 4096, contextWindow: 131072 },
// ],
// },
// 零一万物 (https://platform.lingyiwanwu.com)
// yi: {
// baseUrl: "https://api.lingyiwanwu.com/v1",
// apiKey: "${YI_API_KEY}",
// models: [
// { id: "yi-lightning", maxTokens: 4096, contextWindow: 16384 },
// ],
// },
// MiniMax (https://www.minimaxi.com)
// minimax: {
// baseUrl: "https://api.minimax.chat/v1",
// apiKey: "${MINIMAX_API_KEY}",
// models: [
// { id: "abab6.5s-chat", maxTokens: 2048, contextWindow: 245000 },
// ],
// },
// SiliconFlow 硅基流动 (https://cloud.siliconflow.cn)
// siliconflow: {
// baseUrl: "https://api.siliconflow.cn/v1",
// apiKey: "${SILICONFLOW_API_KEY}",
// models: [
// { id: "deepseek-ai/DeepSeek-V3", maxTokens: 8192, contextWindow: 64000 },
// { id: "Qwen/Qwen2.5-72B-Instruct", maxTokens: 8192, contextWindow: 131072 },
// ],
// },
// 自定义 OpenAI 兼容 API
// custom: {
// api: "openai-completions",
// baseUrl: "https://api.example.com/v1",
// apiKey: "${CUSTOM_API_KEY}",
// },
},
},
// =================================================================
// AGENTS — Agent definitions
// =================================================================
agents: {
defaults: {
compaction: {
mode: "layered", // "default", "safeguard", "layered"
keepRecentPairs: 5,
extractFacts: true,
},
// heartbeat: {
// enabled: true, // Global on/off for heartbeat scanning (default: true)
// }, // Interval/active_hours configured in HEARTBEAT*.md frontmatter
// memory: {
// enabled: true, // Enable per-agent memory (default: true)
// autoCapture: true, // Auto-capture facts from conversations
// autoRecall: true, // Auto-recall relevant memories each turn
// enableManagementTools: true, // Expose memory_search/memory_put/memory_delete tools
// recallTopK: 10, // Per-backend recall count before fusion
// recallFinalK: 5, // Final results after fusion
// // scope: {
// // default: "agent", // Default scope for new memories
// // },
// },
// model: {
// primary: "anthropic/claude-sonnet-4-20250514",
// },
// timeoutSeconds: 600,
// workspace: "~/.rsclaw/workspace",
},
list: [
{
id: "main",
default: true,
model: {
primary: "anthropic/claude-sonnet-4-20250514",
toolset: "full", // "minimal" (6), "standard" (12), "full" (all)
},
workspace: "~/.rsclaw/workspace",
// system: "You are a helpful assistant.",
// channels: [], // Empty = all channels. ["telegram"] = only telegram.
},
// {
// id: "researcher",
// model: { primary: "anthropic/claude-sonnet-4-20250514", toolset: "minimal" },
// system: "You are a research assistant.",
// channels: ["telegram"],
// },
],
// External agents (A2A cross-machine collaboration)
// external: [
// {
// id: "remote_agent",
// url: "http://192.168.0.100:18888",
// authToken: "${REMOTE_AUTH_TOKEN}",
// // remoteAgentId: "main", // Target specific remote agent
// },
// ],
},
// =================================================================
// CHANNELS — Messaging platform connections
// =================================================================
channels: {
// telegram: {
// botToken: "${TELEGRAM_BOT_TOKEN}",
// dmPolicy: "open", // "open", "pairing", "allowlist", "disabled"
// groupPolicy: "allowlist",
// },
// feishu: {
// appId: "cli_xxx",
// appSecret: "${FEISHU_APP_SECRET}",
// brand: "feishu", // "feishu" or "lark"
// connectionMode: "websocket",
// dmPolicy: "pairing",
// groupPolicy: "allowlist",
// },
// dingtalk: {
// appKey: "xxx",
// appSecret: "${DINGTALK_APP_SECRET}",
// dmPolicy: "pairing",
// groupPolicy: "allowlist",
// },
// wechat: {
// botId: "xxx@im.bot",
// botToken: "${WECHAT_BOT_TOKEN}",
// dmPolicy: "pairing",
// groupPolicy: "allowlist",
// },
// wecom: {
// botId: "xxx",
// secret: "${WECOM_SECRET}",
// dmPolicy: "pairing",
// groupPolicy: "open",
// },
// discord: {
// token: "${DISCORD_BOT_TOKEN}",
// dmPolicy: "allowlist",
// },
// slack: {
// botToken: "${SLACK_BOT_TOKEN}",
// appToken: "${SLACK_APP_TOKEN}",
// dmPolicy: "allowlist",
// },
// matrix: {
// homeserver: "https://matrix.org",
// accessToken: "${MATRIX_ACCESS_TOKEN}",
// userId: "@bot:matrix.org",
// deviceId: "xxx",
// },
},
// =================================================================
// TOOLS — Tool configuration
// =================================================================
tools: {
exec: {
safety: false, // Enable exec safety rules
},
upload: {
maxFileSize: 52428800, // 50MB
maxTextChars: 20000,
},
webSearch: {
// provider: "bing-free", // "bing-free", "duckduckgo", "google", "bing", "brave"
// braveApiKey: "${BRAVE_API_KEY}",
// googleApiKey: "${GOOGLE_API_KEY}",
// googleCx: "search-engine-id",
// bingApiKey: "${BING_API_KEY}",
},
// webBrowser: {
// headed: false, // true = visible window, false = headless (default)
// profile: "rsclaw", // Chrome profile: "default" = system profile, "rsclaw" = isolated
// // Unset = temporary profile (no cookies/history)
// // chromePath: "/path/to/chrome", // Auto-detected if not set
// },
},
// =================================================================
// MEMORY — Long-term memory (top-level)
// =================================================================
memory: {
// enabled: true, // Enable memory system (default: true)
// provider: "local", // Embedding provider: "local", "openai", "ollama"
recallTopK: 10, // Results per backend (vector/BM25) before fusion
recallFinalK: 5, // Final results after RRF fusion
// search: {
// model: "bge-small-zh", // Embedding model name
// maxResults: 10, // Max search results
// },
},
// =================================================================
// MEMORY SEARCH — Embedding configuration
// =================================================================
// memorySearch: {
// provider: "local", // "local", "openai", "gemini", "voyage", "ollama"
// model: "bge-small-zh", // Embedding model name
// // sources: ["memory"], // What to index: "memory", "sessions"
// // baseUrl: "http://localhost:11434/v1", // Custom embedding API URL
// // apiKey: "${EMBEDDING_API_KEY}", // API key for cloud embedding
// // local: {
// // modelPath: "~/.rsclaw/models/bge-small-zh", // Path to local GGUF model
// // modelRepo: "BAAI/bge-small-zh-v1.5", // HuggingFace repo name
// // modelDownloadUrl: "https://hf-mirror.com", // Download mirror (auto: hf-mirror.com for zh)
// // },
// },
// =================================================================
// CRON — Scheduled jobs (also see cron/jobs.json)
// =================================================================
// cron: {
// enabled: true,
// maxConcurrentRuns: 2,
// },
// =================================================================
// SESSION — Conversation persistence
// =================================================================
// session: {
// dmScope: "per-channel-peer", // "main", "per-peer", "per-channel-peer", "per-account-channel-peer"
// },
// =================================================================
// LOGGING
// =================================================================
// logging: {
// level: "info", // "trace", "debug", "info", "warn", "error"
// },
}环境变量替换
在任意字符串值中使用 ${VAR_NAME}，加载时自动替换。
json5
{
models: {
providers: {
openai: {
apiKey: "${OPENAI_API_KEY}",
},
},
},
}多实例
bash
rsclaw --dev gateway run # 使用 ~/.rsclaw-dev（端口 18889）
rsclaw --profile test gateway run # 使用 ~/.rsclaw-test提供商自动注册
LLM 提供商从以下来源自动注册：
- 配置文件
models.providers段 - 环境变量（
OPENAI_API_KEY、ANTHROPIC_API_KEY、DASHSCOPE_API_KEY等）
交互式配置
bash
# 完整配置菜单
rsclaw configure
# 直达指定分区
rsclaw configure --section channels
rsclaw configure --section providers
rsclaw configure --section web_search
rsclaw configure --section upload_limits
rsclaw configure --section exec_safety