Mission Control

Loading…

Tasks today

Across all agents

Agents online

Active in last 5 min

Activity (7 days)

Logged events

API spend today

Estimated · Monitor tab

Needs attention

Loading…

Agent workload

Loading…
Fleet activity

Activity by agent

Status distribution

Recent activity

WhenAgentTaskStatusModel

Drag widgets by the ⠿ handle to reorder. Layout is saved in this browser.

Focus session

Focus
25:00

Today's goals

Daily standup

Weekly priorities

Blockers & decisions

Sticky notes

The Team

100%
Auto-refresh 30s

Click a desk to chat · ⓘ for details · working = task in progress · online = active <5 min

Project
Tag

📋 To Do 0

⚙️ In Progress 0

🧑 Needs Owner 0

✅ Done 0

Task

Loading…
Select a file from the tree to view, or pick a folder and upload.

Integrations Hub

Connect OAuth accounts once — agents call APIs through Mission Control proxy.

Connected accounts

Loading…

Available integrations

Curated v1 catalog — more providers added over time.

Pick an agent.
Pick an agent to start chatting.
Type @all or @agentid in this group chat.

API spend

Estimated costs from dashboard chat token logs. Actual provider bills may differ — adjust model rates below.

Today (est.)
This month (est.)
Tokens today
Month budget
Monthly budget

Daily spend (30 days)

Spend by provider

Spend by agent

AgentCallsTokensEst. $

Top costly calls (month)

WhenAgentModelTokEst. $

Spend by chat thread (month)

Find expensive threads — archive or Fresh session to stop context growth.

ThreadCallsTokensEst. $

Org messages

Every inter-agent message, task assign, and strategy rollout step. Failures show what broke and how to fix it. Consult-and-reply uses --reply-thread so answers appear in your chat thread.

WhenFromToKind StatusPreviewTaskThread

Context management

Control transcript caps, context warnings, auto Fresh session, and daily token budgets.

Per-agent context tier

minimal = tight caps · standard = default · full = longer history

Budget & alerts

API keys configured

Loading…

Custom model pricing

Server health
CPU Load (1m)
Memory
Disk
Uptime

Services

Host

Companies

Each company is a separate Mission Control instance. Switch from the header, or add another company below (admin + active license). Local org structure stays under Company. Checkout matches turboanchor.com/pricing (plan, region, company type, billing).

Loading…

Dashboard branding

Header logo and title for the top bar. Upload a separate square favicon for the browser tab (logos are often wide). Leave subtitle blank to hide the tagline pill.

Mission Control

Header logo: PNG, JPG, WebP, GIF, or SVG · max 2 MB · wide images OK

No icon

Square icon ~32×32–512×512 · PNG/ICO recommended

Providers (API keys)

LLM provider keys (OpenAI, Anthropic, …) via OpenClaw. Saving writes to the orchestrator (orchestrator) and syncs to every agent. Keys never echo back — use Replace on a row to rotate a key (same provider/profile).
Loading…
Syncs keys to all agents and restarts the gateway (required for Chat).

Memory embeddings

Controls agents.defaults.memorySearch — the provider OpenClaw uses to embed agent memory for semantic recall. Keys stay in Providers above; this panel only picks provider + model. Changing provider/model pauses vector search until you rebuild indexes. See OpenClaw memory config.

Used when the primary embedding provider fails. Prefer No fallback unless the fallback key is active — a broken OpenAI account can spam retries and overload the gateway.

Loading memory status…

Add agent

Registers a new OpenClaw agent and seeds its workspace (IDENTITY.md, USER.md, memory/). The agent appears in Dashboard, Tasks, and Chat after creation.

Per-agent model

Fleet default: deepseek/deepseek-v4-pro — all agents use this unless you Save a different model below. Overrides persist until you click Reset to default. Model dropdowns show current assignments until you load the catalog.
Loading…

Test a model (round-trip)

Sends a one-shot prompt directly to the model (bypassing agent prompts / skills), so you can isolate whether a provider/key is actually working.
No test yet.

Company

Mission Control’s business deployment layer — identity, who may message whom, and operating rules. Runtime keys stay under AI & keys; member names/photos under Team.

Lowercase id used in Reports to (e.g. ali).

Org structure

Derived from each agent’s IDENTITY.md Reports to field. This is the default communication matrix (before overrides).

Loading…

Communication matrix

Who may message whom via squad-msg / task assign. ✓ = org chart · ★ = allow override · ✕ = deny override.

Loading…

Communication overrides

Rare exceptions on top of the org chart. Stored in team.config.jsoncomm_overrides. Deny wins over org-chart allow; allow opens an otherwise blocked edge.

Loading…

Operating rules

Rule headings in each agent’s USER.md. Apply messaging packs re-injects the clear Messaging + Cost rules (backs up USER.md first) — same as deploy’s apply-rule-messaging-clear.sh.

Loading…

Org traffic

Inter-agent messages, task assigns, and strategy rollouts. Full log lives on Monitor.

Total
Not OK

Stale-task watchdog

Cron (every 30 min) chases assignees on quiet in_progress / assigned todo tasks via squad-msg, escalates to managers, and posts a rollup in the orchestrator chat. Config: team.config.jsonstale_task_watchdog.

Last run: —

Team & org

Names, photos, and titles are saved to each agent's IDENTITY.md (visible in Content tab). Dashboard and Telegram read the same files on every turn. Remove custom agents with Delete (the orchestrator agent cannot be removed). For the company-wide matrix and rule status, open Company in the left nav.
Loading…

Tasks — Projects, work types & tags

Saved to /root/.openclaw/team.config.json under kanban. Projects segregate initiatives (e.g. General, Internal, or your own product names). Work types describe the kind of work (research, development). Tags are separate labels for cross-cutting topics (planning, discussion) — not tied to a single project.

Loading…
Tasks tab refreshes on save. Disabled rows stay on old tasks but won’t appear on new ones.

Channels & pairing

Attach messaging channels (WhatsApp, Telegram) to your orchestrator. For WhatsApp: Show QR to link the phone, then approve each new sender once. Inbound DMs route to your orchestrator agent.

Loading…

Device access

Gateway devices pair with OpenClaw and request scopes. Approve pending upgrades here — operator.admin is required for browser automation and other gateway admin tools. After approval, the local CLI token is refreshed automatically.

Pending requests
Loading…
Paired devices
Loading…

Integration secrets

Tokens and credentials agents read from /root/.openclaw/secrets/ (directory 700, files 600). Use the same filename your agent or script expects (e.g. figma-personal-access-token). Saved values are never shown again — use Replace to paste a new value.
Loading…
Loading…

New support ticket

Select a ticket or create a new one
Loading guide…