Orchestration

Combine enabled skills into a surprising creative showcase on request

Run named agent teams with architect, coder, tester and reviewer roles on Grok Build

Run one autonomous loop pass to check progress, do the next task and reschedule

Predict the best next action from learned patterns and the current state

Coordinate a RuFlo swarm by reading signed updates and checking work claims

Create, assign and update tasks in an AI team run so each has one owner

Schedule persistent background workers that run on a cron-style timer

Join a signed multi-host federation and share work claims across machines

Audit signed RuFlo federation messages and work claims for inconsistencies

Summarize health, channels and active claims on the RuFlo Federation gateway

Build goal-oriented action plans with preconditions, costs and adaptive replanning

Route each task to a model tier using learned patterns and explain the choice

Plan and staff an AI team with roles, a unit budget and bounded tasks

Run background workers on a native loop schedule in your coding tool

Pause, resume or close an AI team run while keeping its evidence

Report AI team progress, blockers, task owners and recorded budget use

Spawn subagents that spawn subagents, up to five levels deep, for delegation

Run one cycle of a sales agent pod against its template and budget

Draft and publish a gateway-attributed RuFlo federation status update

Check the work of a team run against its acceptance criteria

Run ruflo multi-agent swarms with hierarchical coordination across coding tools

Run a swarm agent on a remote ruOS cloud desktop and stream its output back

Start a multi-agent swarm with anti-drift settings for multi-file tasks

Find blockers, collisions and stale work in a RuFlo swarm without changing state

Author a reusable workflow template or an orchestration script with parallel steps

Run, pause, resume and cancel workflows through their full lifecycle

Save a multi-agent task graph by name and re-run it on demand

Run dependency-ordered child tasks as a DAG with retry and amend recovery

Apply the binding ultrawork directive that governs omo execution mode

Create and edit factory.yaml files that define agents, automations and scorers

Coordinate child agent runs with a lead run through the Oz CLI

Report a task as finished or failed back to the Oz platform

Send progress updates to the person who triggered an Oz task, such as via Slack

Run, configure and inspect Oz cloud agents through the Warp REST API and CLI

Report a newly created pull request back to the Oz platform

Upload a local file to the Oz platform as a conversation artifact

Create, hand off and close agent tasks through the MemPalace logstream

Set up recurring marketing workflows that an agent runs on a weekly or daily cadence

Build agent loops with verifiers, stop conditions and proposer-reviewer pairs

Design multi-agent systems and diagnose coordination failures before building

Split compound tasks into research, math, coding and writing stages

Run one iteration of an unattended autonomous development loop

Run roundtable discussions between agent personas or custom expert panels

Delegate coding tasks to the Antigravity CLI non-interactively

Delegate coding tasks to Claude Code and run them non-interactively

Delegate coding tasks to the Codex CLI and run them non-interactively

Delegate coding tasks to DeepSeek Harness and run them non-interactively

Delegate coding tasks to the Gemini CLI and run them non-interactively

Delegate coding tasks to the GitHub Copilot CLI and run them non-interactively