The AgentCore Platform Management Skill (alibabacloud-agentcore-manage) turns AgentCore platform management capabilities into workflows for AI agents. After you install the Skill in an AI assistant, you can use natural language to manage Workspace, Agent, Model, MCP, Skill, AgentSpec, User, Team, and Identity Federation resources.
The Skill includes resource dependencies, operating procedures, parameter contracts, and error-handling instructions to help AI agents complete management tasks in the correct order.
Download
Latest version
(Recommended) For daily use, download the AgentCore Platform Management Skill from the following permanent link. The link always provides the latest released version, so you do not need to change it when the Skill is updated:
Specific version
If you need a fixed version of the Skill, download a versioned package:
Download the AgentCore Platform Management Skill (0.9.15)
The package version aligns with the corresponding AgentCore CLI plug-in version. To download another version, replace 0.9.15 in the file name with the version number of an actual published release.
Install
Extract the ZIP file to a Skill directory supported by your AI assistant. The commands in this section use the global .agents directory (~/.agents/skills) as the installation target. If your AI assistant loads Skills from its own directory, replace ~/.agents/skills in the commands with that directory.
On macOS or Linux, run the following commands:
curl -fL \
'https://agentteams-release-cn-hangzhou.oss-cn-hangzhou.aliyuncs.com/agentcore-manage-skill/alibabacloud-agentcore-manage-latest.zip' \
-o /tmp/alibabacloud-agentcore-manage-latest.zip
mkdir -p ~/.agents/skills
unzip -o /tmp/alibabacloud-agentcore-manage-latest.zip \
-d ~/.agents/skillsAfter extraction, verify that the Skill directory has the following structure:
~/.agents/skills/alibabacloud-agentcore-manage/
├── SKILL.md
├── agents/
└── references/After installation, refresh the Skill or start a new session as required by your AI assistant.
Use
Describe your management goal directly to the AI agent in natural language. For example:
"View AgentCore Workspaces in the Hangzhou region."
"Create an Agent and bind a Model, MCP, and Skill to it."
"Configure Team collaboration for the Workspace."
"Check this Agent's configuration and current status."
For each request, the AI agent follows the instructions in the Skill to check operating conditions, confirm the target environment, identify resource dependencies, perform operations, and read the results.