Unum
model: loading...
session
What would you like to work on?
Choose a starting point or write a direct instruction below.
tokens: --
context: --
speed: --
latency: --
lunum: --
Settings
General
Use the Import button in the chat header to import a session JSON file.
Use the Export button in the chat header to export the current session.
On
On
Order Provider Model Action

Providers

Use compact rows by default. Expand advanced only when you need endpoint details, discovery sources, or deeper diagnostics.

Model Providers

Provider Status Auth Summary Action

Service Providers

Service Status Auth Action

Operator Runtime & Tools

Autonomy

-
workspace

Git Branch

-
No git data

Provider Health

-
No provider telemetry

Context Budget

-
No context telemetry

Phase 0 Diagnostics

-
Runtime state contract not loaded
Config parity not loaded

Autonomy Cycle

-
No cycle snapshot

Self Awareness

-
No self-awareness snapshot

Pending Queue Watchdog

-
No queue diagnostics

Remediation Queue

-
No remediation data

Autonomy Dashboard

Loading autonomy dashboard...

Tactical Ledger

Loading tactical insights...

Tooling and Skills

Tools

Tool Status Class Backend Profiles Local Model Action

Skills

Skill Status Verdict Usage Last Used

Skill Operations

Operation Tool Status Description

Local Model Rollout

Model Installed Allowlist Action

Download Jobs

Job ID Model Status Created Error Action

Browser Control

Browser Health

Checking...
CDP telemetry

CDP Endpoint

-
Awaiting status

Telegram Channel

Activity Center

All scheduled and automated activities in one place.
Loading activities...

LLM Wiki

Persistent knowledge base — the agent scans headlines and loads relevant pages on demand.
Loading...

Proving Ground

The Proving Ground is an isolated sandbox where the agent can safely experiment, self-test, and validate changes before they affect production.

How to use it: Type an experiment goal (e.g. "test that image generation works" or "verify the shell tool produces correct output") and click Start. The agent will run in an isolated copy of your workspace, then report results. Promote (dry-run) evaluates whether the results are safe to promote to production — it shows the decision without making changes.

Safety: Create Snapshot saves a backup of your production database and config before experiments. Restore LKG rolls back to the last known good state if something goes wrong. The kill switch (shown in Survival Status) immediately stops any running experiment.
Survival Status
Proving Ground: unknown
Watchdog: Disabled
Kill Switch: Clear
Last Known Good
No snapshot available
Restore Points
No restore points
Live Activity
Browser Viewport
No active experiment
Experiment

Support Bundle

Send Diagnostics to Unum
Uploads redacted runtime/session diagnostics and recent server log tails for remote debugging.
idle

Mission Runner

idle

Mission Timeline

No active mission.

Scheduled Missions

No scheduled missions.

Automated Tasks

CorpUnum Scout

idle

Scout Status

No scout data.

Control Plane

Backend coverage panel for self-heal, autonomy master, research, operations, and direct API execution.

Self Heal

Autonomy Master

Research

Workers & Self-Edit

Model Scout

Generic Tasks

API Console

idle