Configuration

Data root is controlled by WB_DATA_PATH (default: ~/.weavbot). Config file path is WB_DATA_PATH/config.json. All keys use camelCase (e.g. apiKey, allowFrom). Environment variables override config with the WB_ prefix and __ as the nesting delimiter (e.g. WB_AGENTS__DEFAULTS__MODEL). Set WB_LANG to override CLI language (zh or en) via weavbot.i18n.

Path Description
Config file WB_DATA_PATH/config.json (default ~/.weavbot/config.json)
Workspace WB_DATA_PATH/workspace/ by default
Data directory WB_DATA_PATH/ (default ~/.weavbot/)
Built-in bin ~/.weavbot/bin/ (always in agent tool PATH)
Logs ~/.weavbot/logs/

providers

Object. Key = provider name (e.g. anthropic, openrouter). Each entry:

Key Type Description
mode "openai" | "anthropic" API style; default "openai"
apiKey string API key
apiBase string (optional) Base URL for OpenAI-compatible APIs
extraHeaders object (optional) Extra HTTP headers

Anthropic notes:

  • If the provider endpoint is Anthropic-compatible (not native Anthropic), prefer mode: "openai" first when the endpoint supports OpenAI API style.
  • If you must use mode: "anthropic" on a compatible gateway (for example Bailian Anthropic endpoint), set apiBase to the gateway URL and use debug logging to inspect payload compatibility.
  • Debug env vars for Anthropic mode:
    • WB_DEBUG_ANTHROPIC=1: print redacted request summaries/details.
    • WB_ANTHROPIC_CACHE_CONTROL=0: force-disable cache_control fields for stricter compatibility.

OpenAI-compatible notes:

  • Debug env var for mode: "openai":
    • WB_DEBUG_OPENAI=1: print redacted request summaries/details and failed payload details.

agents.defaults

Key Type Default Description
workspace string workspace Agent workspace path (relative values are resolved from WB_DATA_PATH)
model string — Model name (e.g. claude-sonnet-4-20250514)
provider string — Must match a key in providers
maxTokens int 8192 Max response tokens
temperature float 0.1 Sampling temperature
maxToolIterations int 40 Max tool-call rounds per turn
memoryWindow int 100 Conversation window size
reasoningEffort string | null null "low" | "medium" | "high" for thinking mode

gateway

Key Type Default Description
heartbeat.enabled bool true Enable heartbeat service
heartbeat.intervalS int 1800 Heartbeat interval (seconds)

tools

Key Type Default Description
tools.web.proxy string | null null HTTP/SOCKS5 proxy for web tools
tools.exec.timeout int 60 Shell tool timeout (seconds)
tools.exec.pathAppend string "" Extra PATH for shell; ~/.weavbot/bin is always prepended
tools.restrictToWorkspace bool false Restrict file/shell tools to workspace
tools.mcpServers object {} MCP servers. Key = name. Value: command, args, env (stdio) or url, headers (HTTP); toolTimeout, disabledTools, enabledTools

channels (global)

Key Type Default Description
channels.sendProgress bool true Stream agent text progress to channel
channels.sendToolHints bool true Stream tool-call hints to channel

channels.telegram

Key Type Description
enabled bool Enable channel
token string Bot token from @BotFather
allowFrom string[] Allowed user IDs or usernames; empty = allow all
proxy string | null HTTP/SOCKS5 proxy URL
replyToMessage bool Quote original message in replies

channels.discord

Key Type Description
enabled bool Enable channel
token string Bot token from Discord Developer Portal
allowFrom string[] Allowed user IDs; empty = allow all
gatewayUrl string Discord gateway URL
intents int Gateway intents bitmask

channels.feishu

Key Type Description
enabled bool Enable channel
appId string Feishu Open Platform App ID
appSecret string App Secret
encryptKey string Event subscription encrypt key (optional)
verificationToken string Event verification token (optional)
allowFrom string[] Allowed open_ids; empty = allow all
reactEmoji string Reaction emoji type (e.g. THUMBSUP, OK)

channels.dingtalk

Key Type Description
enabled bool Enable channel
clientId string AppKey
clientSecret string AppSecret
allowFrom string[] Allowed staff_ids; empty = allow all

channels.slack

Key Type Description
enabled bool Enable channel
mode string "socket" supported
webhookPath string Events webhook path
botToken string Bot token (xoxb-…)
appToken string App-level token (xapp-…)
userTokenReadOnly bool Use read-only user token
replyInThread bool Reply in thread
reactEmoji string Reaction emoji (e.g. eyes)
allowFrom string[] Allowed sender user IDs; empty = allow all
groupPolicy string "mention" | "open" | "allowlist"
groupAllowFrom string[] Allowed channel IDs when groupPolicy is allowlist; empty = allow all
dm.enabled bool Enable DMs
dm.policy string "open" | "allowlist"
dm.allowFrom string[] Allowed DM user IDs when policy is allowlist; empty = allow all

channels.qq

Key Type Description
enabled bool Enable channel
appId string Robot AppID from q.qq.com
secret string Robot secret (AppSecret)
allowFrom string[] Allowed openids; empty = allow all

channels.wecom

Key Type Description
enabled bool Enable channel
botId string Bot ID from the Wecom admin console
secret string Long-connection secret
wsUrl string WebSocket endpoint (default wss://openws.work.weixin.qq.com)
heartbeatIntervalSec int Heartbeat interval in seconds
maxMissedPong int Max consecutive missed pong messages before reconnect
reconnectBaseMs int Base reconnect backoff in milliseconds
reconnectMaxMs int Max reconnect backoff in milliseconds
maxReconnectAttempts int Max reconnect attempts; -1 means unlimited retries
requestTimeoutSec int Request timeout in seconds
singleInstanceGuard bool Enable single-instance guard to avoid duplicate connections
allowFrom string[] Allowed user IDs; empty = allow all
perChatPerMinute int Per-chat rate limit per minute
perChatPerHour int Per-chat rate limit per hour
uploadChunkSize int Media upload chunk size in bytes
tempMediaDir string Temporary media directory (relative to workspace)

channels.wechat

Key Type Description
enabled bool Enable channel
accountId string Default account ID (optional when using accounts)
token string Default account token (optional when using accounts)
baseUrl string API base URL (default https://ilinkai.weixin.qq.com)
cdnBaseUrl string CDN base URL (default https://novac2c.cdn.weixin.qq.com/c2c)
routeTag string Optional SKRouteTag header
allowFrom string[] Allowed sender IDs; empty = allow all
requestTimeoutSec int HTTP request timeout in seconds
longPollTimeoutMs int getUpdates long-poll timeout in milliseconds
pollRetryDelayMs int Delay before retry after poll failure
maxConsecutiveFailures int Backoff threshold for consecutive poll errors
sessionPauseMinutes int Pause window when server returns session-expired error (-14)
typingKeepaliveSec int Typing-status keepalive interval hint
tempMediaDir string Temporary media directory (relative to workspace)
stateDir string Persistent state directory for account files and cursors
enabledAccounts string[] Optional allowlist of account keys to run
accounts object Multi-account map. Each account supports: enabled, accountId, token, baseUrl, cdnBaseUrl, routeTag, allowFrom

Notes:

  • weavbot onboard only enables Wechat as a placeholder (channels.wechat.enabled=true); QR login is not executed in onboarding.
  • Run weavbot wechat login to scan QR code and persist account credentials after onboarding.
  • Multi-account runs in one workspace; session keys are scoped by account and peer.

channels.email

Key Type Description
enabled bool Enable channel
consentGranted bool Owner consent to access mailbox
imapHost, imapPort, imapUsername, imapPassword string/int IMAP (receive)
imapMailbox string Mailbox name (default INBOX)
imapUseSsl bool Use SSL for IMAP
smtpHost, smtpPort, smtpUsername, smtpPassword string/int SMTP (send)
smtpUseTls, smtpUseSsl bool SMTP TLS/SSL
fromAddress string From address for replies
autoReplyEnabled bool Send automatic replies
pollIntervalSeconds int Poll interval
markSeen bool Mark messages read
maxBodyChars int Max body length
subjectPrefix string Reply subject prefix
allowFrom string[] Allowed sender addresses; empty = allow all

channels.mochat

Key Type Description
enabled bool Enable channel
baseUrl string Mochat API base URL
socketUrl, socketPath string Socket.IO endpoint
clawToken string Claw token
agentUserId string Agent user ID
sessions, panels string[] Session/panel filters
allowFrom string[] Allowed users; empty = allow all
mention.requireInGroups bool Require @ in groups
groups object Per-group rules (e.g. requireMention)
replyDelayMode string "off" | "non-mention"
replyDelayMs int Delay for non-mention replies
Socket/retry options — socketReconnectDelayMs, socketMaxReconnectDelayMs, socketConnectTimeoutMs, refreshIntervalMs, watchTimeoutMs, watchLimit, retryDelayMs, maxRetryAttempts
CLI Reference Quick Start
This site uses Just the Docs, a documentation theme for Jekyll.