Official agent skill

Cao Session Management

by awslabs in awslabs/cli-agent-orchestrator

Interact with CAO (CLI Agent Orchestrator) — launch multi-agent sessions, check status, send follow-up instructions, unblock stuck terminals, or shut down sessions.

OfficialApache-2.0Auto-check passedAgent Workflows

Install Cao Session Management

skills CLI
$ npx skills add awslabs/cli-agent-orchestrator --skill cao-session-management -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install awslabs/cli-agent-orchestrator cao-session-management --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/awslabs/cli-agent-orchestrator.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/cao-session-management .claude/skills/cao-session-management && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
cao-session-management
GitHub stars
1.4k
Token cost
~2.1k tokens
SKILL.md length
895 words
Files
1
Skills in repo
14
Repo updated
First seen
Licence
Apache-2.0

At a glance

Interact with CAO (CLI Agent Orchestrator) — launch multi-agent sessions, check status, send follow-up instructions, unblock stuck terminals, or shut down sessions.

  • Working with CAO sessions in any capacity
  • SKILL.md covers Overview, Core Concepts, Prerequisites and Discovering Available Profiles, plus 5 more sections
  • Calls curl
  • Tasks that involve Authentication

What it does

Cao Session Management is an agent skill from awslabs/cli-agent-orchestrator, published by the product's own GitHub organization. Interact with CAO (CLI Agent Orchestrator) — launch multi-agent sessions, check status, send follow-up instructions, unblock stuck terminals, or shut down sessions. Use when working with CAO sessions in any capacity.

Its SKILL.md is about 2.1k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Agent Workflows, covering Authentication and Multi-agent orchestration. It works with Amazon Web Services and Model Context Protocol. The repository describes itself as: Multi-agent orchestration for AI coding CLIs — Claude Code, Kiro, Codex, and more, coordinated in isolated tmux sessions. The licence is Apache-2.0.

When your agent uses it

  • Working with CAO sessions in any capacity
  • Tasks that involve Authentication
  • Tasks that involve Multi-agent orchestration

Example prompts

  • “/cao-session-management”

Requirements

  • Python 3

What it can do on your machine

Read from SKILL.md and the folder at commit 089c53c. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • curl

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md. Its commands use curl, which can reach the network depending on how they are called.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Cao Session Management loads about 2.1k tokens when it runs. Until then it costs about 60 tokens; SKILL.md has 895 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~60
When it runs · the whole SKILL.md, loaded when a task matches
~2.1k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from awslabs/cli-agent-orchestrator at commit 089c53c, republished under its Apache-2.0 licence (© awslabs). 895 words, ~2,110 tokens.

Download SKILL.mdSave it as .claude/skills/cao-session-management/SKILL.md (or your agent's skills folder).
name
cao-session-management
description
Interact with CAO (CLI Agent Orchestrator) — launch multi-agent sessions, check status, send follow-up instructions, unblock stuck terminals, or shut down sessions. Use when working with CAO sessions in any capacity.

CAO Session Management

Overview

CAO runs multi-agent workflows in named sessions. A conductor agent inside each session orchestrates the work.

Core Concepts

  • Session: A group of agent terminals working together
  • Conductor: The supervisor terminal — receives instructions, delegates to workers
  • Provider: LLM backend. Default kiro_cli, override with --provider

Prerequisites

Before launching a session, verify:

  • cao-server is running at localhost:9889. Quick check:
    bash
    curl -sf http://localhost:9889/sessions >/dev/null && echo OK || echo "start cao-server"
    If not running, start it in a separate terminal: cao-server.
  • The agent profile is installed. cao launch --agents <profile> fails if the profile is unknown. Install built-ins or custom files with cao install <profile|path|url>.

Discovering Available Profiles

Profiles are CAO-level entities, installed with cao install regardless of which CLI provider runs them. To find available profiles:

SourceCommand
All available profiles across built-in store + local store + provider directoriescurl -sf http://localhost:9889/agents/profiles — canonical, provider-agnostic
Custom/local profile files onlyls ~/.aws/cli-agent-orchestrator/agent-store/
Built-in profiles installed via cao install <name>ls ~/.aws/cli-agent-orchestrator/agent-context/
Profile installation and keyword discoverysee Agent profile installation and profile discovery
Provider-native list (kiro_cli only)kiro-cli agent list — useful because CAO mirrors profiles into ~/.kiro/agents/

The HTTP endpoint is the recommended check: it scans the built-in packaged store, the local store (agent-store/), and provider-specific directories (including agent-context/), then returns a deduplicated list (by profile name, built-in wins) with a source label on each entry.

If unsure which profile to use, ask the user rather than guessing.

Quick Example

A complete, copy-pasteable supervisor launch. The default provider is kiro_cli; pass --provider <name> to use another (claude_code, codex, antigravity_cli, kimi_cli, copilot_cli, opencode_cli, cursor_cli).

This example assumes a configured CAO setup (server running, profiles installed). On an already-configured host you can skip straight to cao launch. The cao install lines below are only for first-time setup; remove them if your CAO is already configured.

bash
# Optional — skip if your CAO is already configured with these profiles.
# Provider-agnostic: `cao install` works for any provider.
cao install code_supervisor
cao install developer
cao install reviewer

# Launch headlessly (assumes cao-server is already running)
cao launch --agents code_supervisor --headless --yolo \
  --session-name my-task --working-directory '/path/to/project' \
  "Build a hello-world Python script. Delegate to developer, then reviewer."

# Same launch on a different provider
# cao launch --agents code_supervisor --provider claude_code --headless --yolo \
#   --session-name my-task --working-directory '/path/to/project' "..."

# Check progress / final output
cao session status cao-my-task
cao session status cao-my-task --workers

# Clean up
cao shutdown --session cao-my-task

Launching a Session

Every cao launch MUST include:

  • --agents PROFILE — see Discovering Available Profiles above; if unclear, ask the user
  • --headless — required from an LLM agent; without it cao tries to attach tmux
  • --session-name NAME — cao adds cao- prefix automatically
  • --working-directory DIR — a wrong path silently breaks the session with no recovery short of shutdown and relaunch. Ask the user if unclear. Always wrap in single quotes to pass the literal path to the server (prevents local shell expansion of ~ or variables before the value reaches cao).
bash
cao launch --agents <profile> --headless --yolo \
  --session-name <name> --working-directory '<path>' "<task>"

--yolo skips confirmation prompts. Required when launching from an agent — interactive prompts will stall the session.

For SOP-driven workflows (Kiro provider): launch with /prompts to discover available SOPs, then send the matched SOP name prefixed with @ (e.g., @my-sop-name), then send the task — each as separate messages after polling for completed status.

Commands

CommandDescription
cao session listList active sessions
cao session status SESSIONConductor status and last response
cao session status SESSION --workersInclude worker terminals
cao session status SESSION --terminal IDDrill into a specific terminal
cao session status SESSION --jsonMachine-readable output; use to extract terminal IDs
cao session send SESSION "msg"Send and wait until completion (sync)
cao session send SESSION "msg" --timeout NSend and wait up to N seconds
cao session send SESSION "msg" --asyncFire-and-forget without waiting
cao session send SESSION "msg" --terminal IDSend to a specific terminal
cao shutdown --session SESSIONShut down a session
cao shutdown --allShut down all sessions

cao session send waits for completion and returns output inline by default. With --async, it sends and returns immediately without waiting. With --timeout N, it waits up to N seconds — if the timeout expires, the agent is still running; check status later. Session names in commands use the cao- prefixed form (e.g. --session-name mywork → use cao-mywork).

A reported status is inferred from the rendered terminal screen, not from a structured protocol, so it can disagree with reality. Before reporting readiness, progress, or completion to a user, corroborate the status with an output read — see cao-session-liveness.

Show full SKILL.md (262 more words)Show less

Worker Communication

Inside a session, the conductor talks to workers via two MCP tools:

Prefer communicating through the conductor (cao session send SESSION "msg") rather than directly to worker terminals. Bypassing the conductor leaves it without state on what was asked or answered, which causes confusion. Two exceptions: unblocking a stuck worker, and follow-up questions to a persistent async worker (see below).

handoff (blocking) — conductor sends task and waits for the worker to reach COMPLETED status, then reads the output. If it times out, the worker is still running — the conductor just stopped waiting.

assign (non-blocking) — conductor sends task and returns immediately. The worker is expected to call send_message back to the conductor's terminal ID when done. Each terminal only knows its own ID via $CAO_TERMINAL_ID.

By default the conductor uses sync (handoff). You can override this by explicitly asking it to use async or sync protocol when sending it a task.

Async workers (assign) stay alive after completing their task and can answer follow-up questions — useful for ongoing investigation where you want to keep querying the same worker. In this case, sending directly to the worker terminal is appropriate:

bash
cao session send SESSION "<follow-up question>" --terminal <worker-terminal-id>

Common Mistakes

Wrong working directory — agents won't find files, builds fail with confusing errors.

Stuck conductor — conductor is waiting on a worker that stopped responding. Check the worker's status first, then decide: prompt it to continue and send results back, or ask it to resend if it already finished. Never re-delegate work that may still be running — it risks duplicate work.

bash
cao session status SESSION --workers
cao session status SESSION --terminal <worker-terminal-id>
cao session send SESSION "continue your work, then send results to terminal <conductor-terminal-id>" --terminal <worker-terminal-id>

Get the conductor's terminal ID from cao session status SESSION --json.

© awslabs, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in skills/cao-session-management of awslabs/cli-agent-orchestrator.

Open the folder on GitHubat commit 089c53c

Compare with similar skills

Cao Session Management next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Cao Session Management compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Cao Session Management this skillawslabs/cli-agent-orchestrator1.4k—~2.1kAutomated safety check: PassApache-2.0
Agent Squad Python Guide2FastLabs/agent-squad7.8k—~4.7kAutomated safety check: PassApache-2.0
AWS Strands Agents Agentcoresammcj/agentic-coding162—~3kAutomated safety check: PassApache-2.0
Codex with ChatGPT Planning LoopXiaoDuoYa/codex-with-chatgpt7.1k—~11kAutomated safety check: NotesMIT
MCP Server Builder with mcp-usemcp-use/mcp-use11k—~923Automated safety check: PassApache-2.0
MemPalace Task HandoffMemPalace/mempalace59k—~1.9kAutomated safety check: PassMIT

Similar skills

  • Agent Squad Python Guide

    2FastLabs/agent-squad

    Map of the agent-squad Python framework for async multi-agent orchestration: which agent, classifier, storage and tool provider to pick, and the pitfalls to avoid.

    7.8k GitHub stars~4.7k tokensUpdated yesterday
    AI & LLM EngineeringAuto-check passed
  • AWS Strands Agents Agentcore

    sammcj/agentic-coding

    A skill your agent uses when working with AWS Strands Agents SDK or Amazon Bedrock AgentCore platform for building AI agents.

    162 GitHub stars~3k tokensUpdated today
    DevOps & CloudAuto-check passed
  • Codex with ChatGPT Planning Loop

    XiaoDuoYa/codex-with-chatgpt

    Uses ChatGPT in the browser as the planning and review brain for a Codex session, with Codex keeping all execution and ChatGPT reading the workspace through a bridge.

    7.1k GitHub stars~11k tokensUpdated 8 days ago
    Agent WorkflowsAuto-check: notes
  • Builds, modifies, debugs, migrates and verifies TypeScript MCP servers and MCP Apps with the mcp-use framework, treating the installed package's types as the source of truth.

    11k GitHub stars~923 tokensUpdated today
    Agent WorkflowsAuto-check passed
  • MemPalace Task Handoff

    MemPalace/mempalace

    Creates, hands off, claims, executes and closes agent tasks through the MemPalace logstream, with approval of the exact task before it is recorded.

    59k GitHub stars~1.9k tokensUpdated 2 days ago
    Agent WorkflowsAuto-check passed
  • Runs a whole project unattended on an agtx kanban board, decomposing the goal, starting tasks, unblocking workers and merging each result.

    1.7k GitHub stars~3.8k tokensUpdated 7 days ago
    Agent WorkflowsAuto-check passed

More from awslabs/cli-agent-orchestrator

All 14 skills in this repo
  • Cao MCP Apps

    awslabs/cli-agent-orchestrator

    Official

    Enable, operate, and extend CAO's MCP Apps surface — the host-rendered fleet dashboard visible inside MCP App hosts (Claude Desktop, ChatGPT, VS Code Copilot, Goose, Postman).

    1.4k GitHub stars~1.9k tokensUpdated today
    Auto-check passed
  • Agui Author

    awslabs/cli-agent-orchestrator

    Official

    Author live dashboard UI from an agent via the emitui MCP tool.

    1.4k GitHub stars~2k tokensUpdated today
    Auto-check passed
  • MCP Apps Builder

    awslabs/cli-agent-orchestrator

    Official

    Load the official MCP Apps builder skills (create-mcp-app, migrate-oai-app, add-app-to-server, convert-web-app) from github.com/modelcontextprotocol/ext-apps.

    1.4k GitHub stars~1.7k tokensUpdated today
    Auto-check passed
  • Cao Plugin

    awslabs/cli-agent-orchestrator

    Official

    Create a new CAO (CLI Agent Orchestrator) plugin. An agent skill from awslabs/cli-agent-orchestrator.

    1.4k GitHub stars~3.1k tokensUpdated today
    Auto-check: notes
  • Cao Provider

    awslabs/cli-agent-orchestrator

    Official

    Create a new CLI agent provider for CAO (CLI Agent Orchestrator).

    1.4k GitHub stars~2.3k tokensUpdated today
    Auto-check passed
  • Cao Agent Routing

    awslabs/cli-agent-orchestrator

    Official

    Find and select the best installed CAO agent profile for a task before delegating with assign or handoff.

    1.4k GitHub stars~552 tokensUpdated today
    Auto-check passed

Categories

Questions about Cao Session Management

What does Cao Session Management do?

Interact with CAO (CLI Agent Orchestrator) — launch multi-agent sessions, check status, send follow-up instructions, unblock stuck terminals, or shut down sessions. Cao Session Management is an agent skill from awslabs/cli-agent-orchestrator, published by the product's own GitHub organization. Interact with CAO (CLI Agent Orchestrator) — launch multi-agent sessions, check status, send follow-up instructions, unblock stuck terminals, or shut down sessions.

When should I use Cao Session Management?

Cao Session Management fits situations like: working with CAO sessions in any capacity; tasks that involve Authentication; tasks that involve Multi-agent orchestration.

How do I install Cao Session Management in Claude Code?

Run `npx skills add awslabs/cli-agent-orchestrator --skill cao-session-management -a claude-code`. Or copy the skill folder (skills/cao-session-management in awslabs/cli-agent-orchestrator) into .claude/skills/cao-session-management in your project. Claude Code loads it when a task matches its description.

How do I install Cao Session Management in Codex?

Run `npx skills add awslabs/cli-agent-orchestrator --skill cao-session-management -a codex`. Or copy the skill folder (skills/cao-session-management in awslabs/cli-agent-orchestrator) into .agents/skills/cao-session-management in your project. Codex loads it when a task matches its description.

Can I use Cao Session Management in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add awslabs/cli-agent-orchestrator --skill cao-session-management -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/cao-session-management, .gemini/skills/cao-session-management, .github/skills/cao-session-management and .opencode/skills/cao-session-management in your project.

What does Cao Session Management need to run?

Going by SKILL.md and its folder, Cao Session Management needs the command-line tools its instructions call (curl). Our summary lists: Python 3.

Does Cao Session Management access the network?

SKILL.md contains no URLs. Its commands use curl, which can reach the network depending on how they are called. This is read from the text; nothing was executed.

Is Cao Session Management safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Cao Session Management use?

Cao Session Management is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Cao Session Management use?

About 2.1k tokens (SKILL.md is roughly 8.4k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Cao Session Management?

Skills that share tags, products or a category with Cao Session Management: Agent Squad Python Guide (2FastLabs/agent-squad, 7.8k stars), AWS Strands Agents Agentcore (sammcj/agentic-coding, 162 stars), Codex with ChatGPT Planning Loop (XiaoDuoYa/codex-with-chatgpt, 7.1k stars) and MCP Server Builder with mcp-use (mcp-use/mcp-use, 11k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Cao Session Management?

awslabs (a GitHub organization, an official publisher) maintains it in awslabs/cli-agent-orchestrator, which has 1,400 GitHub stars. The repository holds 14 skills in this directory. The repository was last updated on October 9, 2026.

Source: awslabs/cli-agent-orchestrator on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.