How do I install the Openclaw Schema Map skill?

Run npx skillmds add nickgallick/openclaw-schema-map in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

What does the Openclaw Schema Map skill do?

Complete map of the OpenClaw config schema from source code. THE single source of truth for valid config keys. It is listed under Coding & Dev Tools on SkillMD.

Is Openclaw Schema Map safe to use?

This skill has not completed SkillMD's automated safety review yet. Capability flags: reads secrets. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Openclaw Schema Map?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Openclaw Schema Map free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Openclaw Schema Map?

nickgallick (@nickgallick) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Openclaw Schema Map by nickgallick — Agent Skill

Name: Openclaw Schema Map
Rating: 4.666666666666667 (3 reviews)
Author: nickgallick

Changelog

2026-03-19: Initial extraction from repos/openclaw/src/config/zod-schema.ts

OpenClaw Config Schema Map (from source code)

SOURCE

Extracted from Zod schema: repos/openclaw/src/config/zod-schema.ts THIS IS AUTHORITATIVE. If docs disagree with this, this is correct.

How to Regenerate

cd /data/.openclaw/workspace-clawexpert/repos/openclaw && git pull
cat src/config/zod-schema.ts

Root-Level Keys (ALL valid keys — schema is `.strict()`)

Key	Type	Required	Notes
`$schema`	string	No	Schema URL
`meta`	object	No	lastTouchedVersion, lastTouchedAt
`env`	object	No	shellEnv, vars, catchall string
`wizard`	object	No	lastRunAt, lastRunVersion, etc.
`diagnostics`	object	No	enabled, flags, stuckSessionWarnMs, otel, cacheTrace
`logging`	object	No	level, file, maxFileBytes, consoleLevel, consoleStyle, redactSensitive, redactPatterns
`cli`	object	No	banner.taglineMode
`update`	object	No	channel (stable/beta/dev), checkOnStart, auto
`browser`	object	No	enabled, cdpUrl, headless, profiles, ssrfPolicy, etc.
`ui`	object	No	seamColor, assistant.name, assistant.avatar
`secrets`	object	No	providers, defaults, resolution
`auth`	object	No	profiles, order, cooldowns
`acp`	object	No	enabled, dispatch, backend, defaultAgent, allowedAgents, stream, runtime
`models`	object	No	mode, providers, bedrockDiscovery
`nodeHost`	object	No	browserProxy
`agents`	object	No	defaults, list[]
`tools`	object	No	web, media, links, sessions, exec, fs, etc.
`bindings`	array	No	route and acp bindings
`broadcast`	object	No	agent-to-agent broadcast config
`audio`	object	No	audio config
`media`	object	No	preserveFilenames, ttlHours
`messages`	object	No	MessagesSchema
`commands`	object	No	CommandsSchema
`approvals`	object	No	ApprovalsSchema
`session`	object	No	SessionSchema
`cron`	object	No	enabled, store, maxConcurrentRuns, retry, webhook, sessionRetention, runLog, failureAlert, failureDestination
`hooks`	object	No	enabled, path, token, mappings, gmail, internal, etc.
`web`	object	No	enabled, heartbeatSeconds, reconnect
`channels`	object	No	ChannelsSchema (all channel types)
`discovery`	object	No	wideArea, mdns
`canvasHost`	object	No	enabled, root, port, liveReload
`talk`	object	No	provider, providers, voiceId, modelId, apiKey, etc.
`gateway`	object	No	port, mode, bind, customBindHost, controlUi, auth, tls, http, etc.
`memory`	object	No	backend, citations, qmd
`mcp`	object	No	servers record (stdio MCP servers)
`skills`	object	No	allowBundled, load, install, limits, entries
`plugins`	object	No	enabled, allow, deny, load, slots, entries, installs

REJECTED/INVALID ROOT KEYS: Any key not in the above list will crash the gateway (.strict() mode).

`agents` schema

`agents.defaults`

See AgentDefaultsSchema — same shape as agent entries but all optional.

`agents.list[]` item (AgentEntrySchema)

Key	Type	Required	Notes
`id`	string	YES	Unique agent identifier
`default`	boolean	No	Is this the default agent?
`name`	string	No	Display name
`workspace`	string	No	Path to workspace directory
`agentDir`	string	No	Alternative: agent directory path
`model`	AgentModelSchema	No	Model string or object
`skills`	string[]	No	Skill IDs to load
`memorySearch`	MemorySearchSchema	No	Memory search config
`humanDelay`	HumanDelaySchema	No	mode, minMs, maxMs
`heartbeat`	HeartbeatSchema	No	every, activeHours, model, prompt, etc.
`identity`	IdentitySchema	No	name, theme, emoji, avatar
`groupChat`	GroupChatSchema	No	mentionPatterns, historyLimit
`subagents`	object	No	allowAgents, model, thinking
`sandbox`	AgentSandboxSchema	No	docker, network, etc.
`params`	record	No	Arbitrary key-value params
`tools`	AgentToolsSchema	No	Per-agent tool overrides
`runtime`	AgentRuntimeSchema	No	Runtime config

`gateway` schema

Key	Type	Notes
`port`	int	Gateway port
`mode`	"local" \| "remote"
`bind`	"auto" \| "lan" \| "loopback" \| "custom" \| "tailnet"
`customBindHost`	string	VALID — use with bind:"custom"
`controlUi`	object	enabled, basePath, allowedOrigins, etc.
`auth`	object	mode, token, password, rateLimit, trustedProxy
`auth.mode`	"none" \| "token" \| "password" \| "trusted-proxy"
`tls`	object	enabled, autoGenerate, certPath, keyPath
`tailscale`	object	mode (off/serve/funnel), resetOnExit
`remote`	object	url, transport, token, sshTarget, etc.
`reload`	object	mode (off/restart/hot/hybrid), debounceMs
`http`	object	endpoints (chatCompletions, responses), securityHeaders
`push`	object	apns relay config
`nodes`	object	browser mode/node, allowCommands, denyCommands
`tools`	object	deny, allow arrays
`channelHealthCheckMinutes`	int	0=disabled, default 5
`channelStaleEventThresholdMinutes`	int	Must be >= healthCheckMinutes
`channelMaxRestartsPerHour`	int
`trustedProxies`	string[]
`allowRealIpFallback`	boolean

`channels` schema

Defined in zod-schema.providers.ts — includes: telegram, discord, slack, signal, whatsapp, imessage, irc, msteams, googlechat, feishu, webchat, and more.

`cron` schema

Key	Type	Notes
`enabled`	boolean
`store`	string	Storage path
`maxConcurrentRuns`	int
`retry`	object	maxAttempts, backoffMs[], retryOn[]
`retry.retryOn`	enum[]	rate_limit, overloaded, network, timeout, server_error
`webhook`	http url
`webhookToken`	SecretInput
`sessionRetention`	string \| false	Duration string (ms/s/m/h/d) or false
`runLog`	object	maxBytes (string/number), keepLines
`failureAlert`	object	enabled, after, cooldownMs, mode, accountId
`failureDestination`	object	channel, to, accountId, mode

`mcp` schema (NATIVE stdio MCP — not mcporter)

mcp:
  servers:
    <server-name>:
      command: string       # stdio server command
      args: string[]
      env: record
      cwd: string
      workingDirectory: string
      url: http url         # for HTTP-based servers

ACP capabilities (http/sse) are false — but stdio MCP works via mcp.servers.

`memory` schema

Key	Type	Notes
`backend`	"builtin" \| "qmd"
`citations`	"auto" \| "on" \| "off"
`qmd`	object	command, mcporter, searchMode, paths, sessions, update, limits, scope
`qmd.searchMode`	"query" \| "search" \| "vsearch"
`qmd.mcporter.enabled`	boolean	Route QMD through mcporter daemon

`skills` schema

Key	Type	Notes
`allowBundled`	string[]
`load.extraDirs`	string[]	Additional skill dirs
`load.watch`	boolean
`install.preferBrew`	boolean
`install.nodeManager`	npm/pnpm/yarn/bun
`limits`	object	maxCandidatesPerRoot, maxSkillsLoadedPerSource, etc.
`entries.<id>`	SkillEntrySchema	enabled, apiKey, env, config

`plugins` schema

Key	Type	Notes
`enabled`	boolean
`allow`	string[]
`deny`	string[]
`load.paths`	string[]
`slots.memory`	string	Plugin ID for memory slot
`slots.contextEngine`	string	Plugin ID for context engine slot
`entries.<id>`	PluginEntrySchema	enabled, hooks, subagent, config
`installs.<id>`	PluginInstallRecordShape

`secrets` schema

Key	Type	Notes
`providers.<name>`	SecretProviderSchema	source: env/file/exec
`defaults.env`	string	Default env provider alias
`defaults.file`	string	Default file provider alias
`defaults.exec`	string	Default exec provider alias
`resolution`	object	maxProviderConcurrency, maxRefsPerProvider, maxBatchBytes

SecretInput type accepts: plain string OR { source: "env"/"file"/"exec", provider: string, id: string }

`auth` schema

Key	Type	Notes
`profiles.<name>`	object	provider, mode (api_key/oauth/token), email
`order`	record	provider → auth profile order
`cooldowns`	object	billingBackoffHours, failureWindowHours, etc.

`logging` schema

Key	Type	Notes
`level`	silent/fatal/error/warn/info/debug/trace
`file`	string	Log file path
`maxFileBytes`	int
`consoleLevel`	same enum
`consoleStyle`	pretty/compact/json
`redactSensitive`	"off" \| "tools"
`redactPatterns`	string[]

`update` schema

Key	Type	Notes
`channel`	"stable" \| "beta" \| "dev"
`checkOnStart`	boolean
`auto.enabled`	boolean
`auto.stableDelayHours`	number	Max 168
`auto.betaCheckIntervalHours`	number	Max 24

`acp` schema

Key	Type	Notes
`enabled`	boolean
`dispatch.enabled`	boolean
`backend`	string
`defaultAgent`	string
`allowedAgents`	string[]
`maxConcurrentSessions`	int
`stream`	object	coalesceIdleMs, maxChunkChars, deliveryMode, etc.
`runtime.ttlMinutes`	int
`runtime.installCommand`	string

`diagnostics.otel` schema

Key	Type	Notes
`enabled`	boolean
`endpoint`	string
`protocol`	"http/protobuf" \| "grpc"
`headers`	record
`traces/metrics/logs`	boolean
`sampleRate`	0-1
`flushIntervalMs`	int

Rejected / Invalid Root Keys

mcpServers — NEVER a root key; use mcp.servers instead
Any unknown key at root level → Zod .strict() throws → gateway crash

Openclaw Schema Map

File contents

Changelog

OpenClaw Config Schema Map (from source code)

SOURCE

How to Regenerate

Root-Level Keys (ALL valid keys — schema is `.strict()`)

`agents` schema

`agents.defaults`

`agents.list[]` item (AgentEntrySchema)

`gateway` schema

`channels` schema

`cron` schema

`mcp` schema (NATIVE stdio MCP — not mcporter)

`memory` schema

`skills` schema

`plugins` schema

`secrets` schema

`auth` schema

`logging` schema

`update` schema

`acp` schema

`diagnostics.otel` schema

Rejected / Invalid Root Keys

Frequently asked questions

Openclaw Schema Map

File contents

Changelog

OpenClaw Config Schema Map (from source code)

SOURCE

How to Regenerate

Root-Level Keys (ALL valid keys — schema is .strict())

agents schema

agents.defaults

agents.list[] item (AgentEntrySchema)

gateway schema

channels schema

cron schema

mcp schema (NATIVE stdio MCP — not mcporter)

memory schema

skills schema

plugins schema

secrets schema

auth schema

logging schema

update schema

acp schema

diagnostics.otel schema

Rejected / Invalid Root Keys

Related skills

Frequently asked questions

How do I install the Openclaw Schema Map skill?

What does the Openclaw Schema Map skill do?

Is Openclaw Schema Map safe to use?

Which AI agents work with Openclaw Schema Map?

Is Openclaw Schema Map free to use?

Who published Openclaw Schema Map?

Root-Level Keys (ALL valid keys — schema is `.strict()`)

`agents` schema

`agents.defaults`

`agents.list[]` item (AgentEntrySchema)

`gateway` schema

`channels` schema

`cron` schema

`mcp` schema (NATIVE stdio MCP — not mcporter)

`memory` schema

`skills` schema

`plugins` schema

`secrets` schema

`auth` schema

`logging` schema

`update` schema

`acp` schema

`diagnostics.otel` schema