openclaw/docs/channels/clickclack.md

summary, read_when, title

summary	read_when	title
ClickClack bot-token channel setup and target syntax	Connecting OpenClaw to a ClickClack workspace Testing ClickClack bot identities	ClickClack

ClickClack connects OpenClaw to a self-hosted ClickClack workspace through first-class ClickClack bot tokens.

Use this when you want an OpenClaw agent to appear as a ClickClack bot user. ClickClack supports independent service bots and user-owned bots; user-owned bots keep an owner_user_id and receive only the token scopes you grant.

Quick setup

In ClickClack, open Workspace settings → Integrations → OpenClaw, create a bot using Setup code (recommended), and copy the generated command:

openclaw channels add clickclack --code 'https://clickclack.example.com/#XXXX-XXXX-XXXX'

For separate frontend and API origins or a path-mounted API, ClickClack emits an exact claim endpoint instead:

openclaw channels add clickclack --code 'https://api.example.com/services/clickclack/api/bot-setup-codes/claim#XXXX-XXXX-XXXX'

The setup code is single-use and expires after 10 minutes. OpenClaw claims it, receives the newly minted bot token and workspace settings, saves the account, verifies the connection, and reports whether the running gateway picked it up. For versioned exact endpoints, OpenClaw validates and saves the canonical API base returned by ClickClack, including any path prefix. The setup code itself is not stored in OpenClaw config.

Setup-code claims use HTTPS for public servers. Plain HTTP is also supported for local installations on loopback addresses such as localhost and 127.0.0.1.

If OpenClaw is already running, ClickClack connects automatically and no second command is needed. Otherwise, start it with:

openclaw gateway

You can also pass the code separately from the server URL:

openclaw channels add clickclack --code XXXX-XXXX-XXXX --base-url https://clickclack.example.com

For guided setup, run:

openclaw onboard

Select ClickClack, then enter the server URL, bot token, and workspace when prompted. Guided setup checks the server, token, and workspace after saving; a failed check does not discard the configuration.

Alternative: manual token

Choose Manual token in ClickClack when configuring a non-OpenClaw client or when you explicitly need to manage the token yourself:

openclaw channels add clickclack --base-url https://clickclack.example.com --token ccb_... --workspace default

workspace accepts a workspace id (wsp_...), slug, or display name. --code cannot be combined with --token, --token-file, or --use-env.

Alternative: env-based token

The default account can read CLICKCLACK_BOT_TOKEN instead of storing a token in config:

export CLICKCLACK_BOT_TOKEN="ccb_..."
openclaw channels add clickclack --base-url https://clickclack.example.com --workspace default --use-env
openclaw gateway

Named accounts must use a configured token or token file; the shared env variable is intentionally limited to the default account.

JSON5 reference

The equivalent config shape is:

{
  channels: {
    clickclack: {
      enabled: true,
      baseUrl: "https://clickclack.example.com",
      token: { source: "env", provider: "default", id: "CLICKCLACK_BOT_TOKEN" },
      workspace: "default",
      defaultTo: "channel:general",
    },
  },
}

An account counts as configured only when baseUrl, a token source, and workspace are all set. A token source can be token, tokenFile, or CLICKCLACK_BOT_TOKEN for the default account. workspace accepts a workspace id (wsp_...), slug, or name; the gateway resolves it to the id at startup.

Account config keys

Key	Default	Notes
baseUrl	none (required)	Public ClickClack URL used for browser-facing links.
apiBaseUrl	baseUrl	Optional server-to-server endpoint for REST and realtime WebSocket traffic.
token	none	Bot token as a plain string or secret ref (source: "env" | "file" | "exec").
tokenFile	none	Path to a bot-token file; takes precedence over token.
workspace	none (required)	Workspace id, slug, or name.
replyMode	"agent"	"agent" runs the full agent pipeline; "model" sends short direct model completions.
defaultTo	"channel:general"	Target used when an outbound path gives no target.
allowFrom	["*"]	User-id allowlist for inbound DMs and channel messages.
botUserId	auto-detected	Resolved from the bot token identity at startup.
agentId	route default	Pin this account's inbound messages to one agent.
toolsAllow	none	Tool allowlist for agent replies from this account.
model, systemPrompt	none	Used by replyMode: "model" completions.
commandMenu	true	Publish native commands to ClickClack composer autocomplete.
reconnectMs	1500	Realtime reconnect delay (100 to 60000).
discussions	disabled	Managed per-session channel settings; see Session discussions.

Keep an auth-gated public hostname

Use apiBaseUrl when ClickClack and the OpenClaw gateway run on the same host but the public ClickClack hostname is protected by an authentication gateway such as Cloudflare Access:

{
  channels: {
    clickclack: {
      baseUrl: "https://clack.openclaw.ai",
      apiBaseUrl: "http://127.0.0.1:8484",
      token: { source: "env", provider: "default", id: "CLICKCLACK_BOT_TOKEN" },
      workspace: "default",
    },
  },
}

The public hostname can remain fully auth-gated for browser users. OpenClaw uses the loopback endpoint for REST requests, setup verification, and the realtime WebSocket, while discussion embedUrl and openUrl links continue to use the public baseUrl. If apiBaseUrl is omitted, all traffic uses baseUrl, preserving existing behavior.

If plugins.allow is a non-empty restrictive list, explicitly selecting ClickClack in channel setup or running openclaw plugins enable clickclack appends clickclack to that list. Onboarding installation uses the same explicit-selection behavior. These paths do not override plugins.deny or a global plugins.enabled: false setting. Direct openclaw plugins install @openclaw/clickclack follows the normal plugin-install policy and also records ClickClack in an existing allowlist.

Multiple bots

Each account opens its own ClickClack realtime connection and uses its own bot token.

{
  channels: {
    clickclack: {
      enabled: true,
      baseUrl: "https://clickclack.example.com",
      defaultAccount: "service",
      accounts: {
        service: {
          token: { source: "env", provider: "default", id: "CLICKCLACK_SERVICE_BOT_TOKEN" },
          workspace: "default",
          defaultTo: "channel:general",
          agentId: "service-bot",
        },
        support: {
          token: { source: "env", provider: "default", id: "CLICKCLACK_SUPPORT_BOT_TOKEN" },
          workspace: "default",
          defaultTo: "dm:usr_...",
          agentId: "support-bot",
        },
      },
    },
  },
}

Session discussions

Enable discussions on one ClickClack account to give each OpenClaw session a dedicated ClickClack channel. The account token must include channels:write (the bot:admin bundle includes it); the normal bot:write setup token cannot create or synchronize channels.

{
  channels: {
    clickclack: {
      enabled: true,
      baseUrl: "https://clickclack.example.com",
      token: { source: "env", provider: "default", id: "CLICKCLACK_BOT_TOKEN" },
      workspace: "default",
      discussions: {
        enabled: true,
        workspace: "default",
        controlUrlBase: "https://team.openclaw.ai",
        section: "Sessions",
      },
    },
  },
}

discussions.workspace accepts the same workspace id, slug, or display name as the account-level workspace and defaults to that value. section controls the ClickClack sidebar section and defaults to Sessions. When controlUrlBase is set, the managed channel links back to the real Control UI session route, /chat?session=<encoded-session-key>.

Enable discussions on exactly one ClickClack account. The gateway provider has no account selector, so multiple enabled discussion accounts are rejected rather than choosing one by configuration order.

Opening a discussion creates a public ClickClack channel marked as externally managed. The plugin keeps the session label, category, and archive state in sync. Restoring a session restores its channel; clearing the session category moves the channel back to the configured default section. Deleting an OpenClaw session archives the ClickClack channel instead of deleting it, so its history remains available. The plugin reconciles bindings when discussion RPCs are used and approximately once per minute while any bindings exist.

Inbound messages in a managed channel use a deterministic side session under the same agent id as the attached main session. The side agent is told which main session to observe and can use sessions_history and session_status (changesSince is useful for incremental checks). It uses sessions_send only when people in the discussion ask it to relay or steer the main session. The binding, managed ownership reference, and side-session peer identity include the concrete OpenClaw session id along with the pinned ClickClack server and channel. Resetting a reusable session key or retargeting an account revokes the old channel locally, archives it when the old credential remains usable, and cannot reuse its side transcript. Messages arriving through an archived, reset, disabled, or retargeted binding are dropped instead of falling back to the account's normal channel routing. Released bindings leave a durable revoked-channel marker so delayed realtime events remain fail-closed. Remote ownership is keyed by ClickClack server and channel id, so renaming the local account cannot turn a managed channel into an ordinary one.

Keep tools.sessions.visibility at its safer default tree. The plugin installs a host-scoped grant only between each side session and its attached main session, plus a tool-policy hook that blocks session discovery and cross-session targets. It allows sessions_history, session_status, and sessions_send only for the attached main session and prevents the status call from changing that session's model. Those tools must still be present in the agent's effective tool allowlist. The system prompt is guidance; the host grant and hook are the authorization boundary.

The ClickClack server must support managed-channel fields (external_managed, external_ref, external_url, and sidebar_section) on channel creation and updates and return them in channel responses. OpenClaw verifies that contract before persisting a binding. If a create response is lost, the next open adopts the channel by its server-enforced external_ref instead of creating another. Until that outcome is reconciled, the pending reservation quarantines otherwise-unbound events in the destination workspace. The coarse reconciler adopts the channel when the same session is still live or archives it after a reset; it clears the reservation when no remote channel was created. That reference contains a durable per-OpenClaw-installation namespace plus a hash of the session key, concrete session id, ClickClack destination, and durable binding generation. Separate gateways cannot adopt each other's channels, reset sessions cannot inherit old channel history, and an account or workspace round trip cannot re-adopt a previous channel. Bindings are also pinned to the configured ClickClack server URL and are invalidated if the account is retargeted. Changing or removing controlUrlBase updates or clears the managed channel link on the next reconciliation pass. Changing discussions.workspace archives and releases the old binding before a channel can be opened in the new workspace when the old workspace credential remains configured. If the token was replaced with a workspace-scoped credential that cannot access the old workspace, OpenClaw records the old channel as revoked and releases the binding without trying the replacement token; archive that leftover channel from ClickClack.

The attached main session also receives a pull-only discussion tool. It reads the latest messages and recent thread replies as one escaped, attributed record per message, and has no write or lifecycle side effects. Channel-root and thread lookups have fixed request budgets; the result explicitly warns when that safety bound can omit an older active thread.

Reply modes

• replyMode: "agent" (default) dispatches inbound messages through the normal agent pipeline, including session recording and tool policy.
• replyMode: "model" skips the agent pipeline and uses the plugin runtime's llm.complete for direct bot replies, optionally shaped by model and systemPrompt. The selected provider and model own the completion budget.

Model mode runs completions against the resolved bot agent id, which requires the explicit plugins.entries.clickclack.llm.allowAgentIdOverride: true trust bit:

{
  plugins: {
    entries: {
      clickclack: {
        llm: {
          allowAgentIdOverride: true,
        },
      },
    },
  },
}

Keep the trust bit off if you only use the default agent reply mode; it is not needed there.

Command menu

At gateway startup, each configured account publishes OpenClaw's native commands to ClickClack. They appear in composer autocomplete labeled with the bot's handle. The published set is replaced wholesale on each startup, including clearing a stale menu when the native command catalog is empty.

Command-menu sync is enabled by default. Set commandMenu: false on an account to opt out:

{
  channels: {
    clickclack: {
      enabled: true,
      token: { source: "env", provider: "default", id: "CLICKCLACK_BOT_TOKEN" },
      workspace: "default",
      commandMenu: false,
    },
  },
}

The token needs commands:write. Current ClickClack bot:write and bot:admin bundles include that scope, and it can also be granted individually. Tokens created before command menus were introduced may need the scope added or a replacement token.

Sync is best effort and runs once per gateway start. A missing scope or network failure logs a warning; an older ClickClack server without the endpoint logs at debug level. None of these failures block realtime startup. Menus remain available while the agent is offline and are removed when the bot leaves the workspace.

This release publishes native command specs only. Aliases and skill-, plugin-, or custom-command catalogs are not added to the menu. If a name is also registered as an HTTP slash command, ClickClack dispatches that registration first; other menu commands continue through normal message delivery.

Use agent mode for cross-service correlation evidence. For an authoritative ClickClack message id in its canonical msg_<ulid> shape, the channel derives the deterministic OpenClaw run id clickclack:<message-id>. Each model call is then visible in diagnostics as clickclack:<message-id>:model:<n>; when that turn uses ClawRouter, the same model-call id is sent as X-Request-ID. model mode bypasses the normal agent run/session diagnostics and is therefore not suitable for this evidence path.

When a realtime event contains a validated payload.correlation_id, the channel carries it as X-Correlation-ID on the authoritative message fetch and the resulting ClickClack reply requests. Values use ClickClack's safe 128-character set (A-Z, a-z, 0-9, ., _, :, and -); invalid values are omitted. These joins contain identifiers only, never message bodies, prompts, completions, credentials, or tool output.

Durable media delivery

Agent replies containing media use required durable delivery. OpenClaw assigns stable per-part message and upload nonces before the first ClickClack write, so a retry reuses the same upload and message instead of consuming storage quota or publishing duplicates. If an upload already exists after a restart, OpenClaw does not reread the original local path or remote media URL.

This recovery contract requires a ClickClack server that supports:

• GET /api/uploads/by-nonce with X-ClickClack-Upload-Nonce: supported on found and missing results.
• GET /api/messages/by-nonce with X-ClickClack-Message-Nonce: supported on found and missing results.
• Idempotent message creation and attachment association for the same owner-scoped nonce and upload.

An older server's generic 404 is not treated as proof that a send is absent. OpenClaw leaves the delivery unresolved rather than risking a duplicate; update ClickClack before enabling media-producing agent replies.

Agent activity rows

By default a ClickClack channel shows nothing while an agent turn runs; only the final reply lands. Set agentActivity: true on an account to publish durable agent_commentary and agent_tool message rows while the turn is in progress:

{
  channels: {
    clickclack: {
      enabled: true,
      token: { source: "env", provider: "default", id: "CLICKCLACK_BOT_TOKEN" },
      workspace: "default",
      agentActivity: true,
    },
  },
}

Requirements and behavior:

• Off by default. Stock setups and older ClickClack servers are untouched.
• Requires the agent_activity:write token scope. This scope is separate from bot:write and is not inherited by it; create the bot token with --scopes bot:write,agent_activity:write (or grant the scope to an existing token) before enabling the option.
• Best-effort degradation. If the token lacks agent_activity:write or the server rejects activity writes, failures are logged and the final reply still delivers normally; no activity rows appear.
• Rows are grouped per turn (turn_id), coalesced so one logical step is one row, and tool rows use the same progress formatting as Discord/Slack/Telegram (tool name plus command detail).
• Attribution metadata. Agent-authored posts (activity rows and the final reply) carry author_model and author_thinking fields resolved from the actual model used for the turn (including after fallback). Servers that do not define these columns ignore the unknown JSON fields; servers that persist them can answer "which model said this line, at which thinking level" per message.

Targets

• channel:<name-or-id> sends to a workspace channel. Bare targets default to channel:.
• dm:<user_id> creates or reuses a direct conversation with that user.
• thread:<message_id> replies in the thread rooted at that message.

Explicit outbound targets may also carry the clickclack: or cc: provider prefix.

Outbound media uses ClickClack's upload API and then attaches the durable upload to the created channel message, thread reply, or DM. Local files and supported remote media URLs follow OpenClaw's normal media-access policy, with a 64 MiB per-file limit. Durable queued sends use separate owner-scoped nonces for each upload and message part, then retry attachment association with those same objects. See Durable media delivery for the server contract and recovery behavior.

Examples:

openclaw message send --channel clickclack --target channel:general --message "hello"
openclaw message send --channel clickclack --target dm:usr_123 --message "hello"
openclaw message send --channel clickclack --target thread:msg_123 --message "following up"

Permissions

ClickClack token scopes are enforced by the ClickClack API.

• bot:read: read workspace/channel/message/thread/DM/realtime/profile data.
• bot:write: bot:read plus channel messages, thread replies, DMs, uploads, and command-menu publishing.
• bot:admin: bot:write plus channel creation.
• commands:write: publish the bot's command menu. Included in current bot:write and bot:admin bundles and grantable individually.
• agent_activity:write: durable agent activity rows (agent_commentary / agent_tool). Not inherited by bot:write or bot:admin; required only when agentActivity: true is set.

OpenClaw only needs current bot:write for normal agent chat and command-menu sync. Add agent_activity:write when enabling agent activity rows.

Troubleshooting

• ClickClack is not configured for account "<id>": set baseUrl, token (for example via CLICKCLACK_BOT_TOKEN), and workspace for that account.
• ClickClack workspace not found: <value>: set workspace to the workspace id, slug, or name returned by ClickClack.
• No inbound replies: confirm the token has realtime read access and note that the bot ignores its own messages and messages from other bots.
• Channel sends fail: verify the bot is a member of the workspace and has bot:write.
• No command menu: confirm commandMenu is not false, the ClickClack server supports PUT /api/bots/self/commands, and the token has commands:write.