Skip to content
Back to skills

Using Ao

ASecurity

Catalog of the AO Cloud `ao` CLI available inside a cloud sandbox: spawning worker sessions, listing them, messaging them, reporting to the orchestrator, killing them, and claiming pull requests. Use when orchestrating or coordinating AO cloud sessions with the ao CLI.

  • 12,858 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added October 5, 2026
ai-agents

Works with

  • cli

Security analysis

A100/100

Pro scans all 3 files and shows the line behind each finding

Scanned October 5, 2026

npx -y skills add OrchestratorInc/agent-orchestrator --skill using-ao --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Using Ao?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Using Ao
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/orchestratorinc-using-ao-agent-orchestrator/badge)](https://www.skillsdirectory.com/skills/orchestratorinc-using-ao-agent-orchestrator)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: using-ao
description: "Catalog of the AO Cloud `ao` CLI available inside a cloud sandbox: spawning worker sessions, listing them, messaging them, reporting to the orchestrator, killing them, and claiming pull requests. Use when orchestrating or coordinating AO cloud sessions with the ao CLI."
trigger: "Using the ao CLI inside an AO Cloud sandbox: spawning, listing, messaging, reporting to, or killing sessions, or claiming a PR."
---

# AO Cloud CLI Catalog

`ao` inside a cloud sandbox is a thin client over the AO control plane. It is a
different, smaller CLI than the desktop `ao`: there is no `ao session`,
`ao project`, `ao status`, `ao preview`, or `ao browser` here. Every command is
`ao help` for the authoritative list.

| Command | What it does | Who can run it | Details |
|---|---|---|---|
| `spawn` | Create a worker session in its own fresh sandbox | Orchestrators only | [commands/orchestration.md](commands/orchestration.md) |
| `list` (`ls`, `status`) | List the workers this orchestrator spawned, with branch, status, and PR facts | Orchestrators only | [commands/orchestration.md](commands/orchestration.md) |
| `send` | Send a message into a worker's conversation | Orchestrators only | [commands/orchestration.md](commands/orchestration.md) |
| `kill` (`delete`, `rm`) | Terminate a worker session and its sandbox | Orchestrators only | [commands/orchestration.md](commands/orchestration.md) |
| `report` | Send a message to the orchestrator that spawned this session | Workers with an orchestrator parent | [commands/orchestration.md](commands/orchestration.md) |
| `claim-pr` | Attach an existing pull request to this session | Any session | [commands/pull-requests.md](commands/pull-requests.md) |

## Conventions

- `ao list --json` prints machine-readable output; `--all` includes terminated
  workers (hidden by default).
- `--agent` is an alias for `--harness` on `ao spawn`.
- Session ids are UUIDs shown by `ao list`.
- Every session runs in its own isolated sandbox. Never try to reach another
  session's sandbox directly (no ssh, no shared filesystem); the `ao` commands
  above are the only channel between sessions.
- A `send` to a worker whose sandbox is still provisioning is queued by the
  control plane and delivered when the worker's agent is ready — check
  `runtimeConnected` in `ao list --json` to see whether a worker is live.
- Spawning and messaging are idempotent server-side; a command that prints a
  success line has been durably accepted even if the effect is not visible yet.
- Pull requests and reviews are handled through socket helpers described in
  [commands/pull-requests.md](commands/pull-requests.md), not through `ao`.

Files in this skill

  • SKILL.md2.6 KB
  • commands/orchestration.md4 KB
  • commands/pull-requests.md1.3 KB

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.