Automation and AI
Overview
Use the AGS CLI in CI pipelines and with AI assistants.
CI pipelines
Use --no-input to suppress interactive prompts in CI environments. Use --format json to consume command output in scripts.
Authenticate with the client-credentials grant using a confidential IAM client:
ags auth login --grant client-credentials
Store the client secret as a CI secret variable. Do not hardcode it in scripts or workflow files.
AI assistants
Run ags describe to give an AI agent the full command surface. The agent selects a workflow or command and fills in the required inputs. The CLI runtime executes deterministically: the same command always calls the same endpoint with the same parameters. The agent cannot cause unexpected side effects by choosing a command.
The AI plugin
The AI Marketplace Plugin at github.com/AccelByte/ai-plugins connects AI coding assistants such as Claude Code and Codex to the AGS CLI. It exposes AGS workflows as callable tools within the assistant, so the agent can invoke them directly without constructing raw CLI commands.
MCP servers
The AGS API MCP Server at github.com/AccelByte/ags-api-mcp-server gives an AI assistant direct access to AGS API endpoints and schemas.
The AGS Extend SDK MCP Server at github.com/AccelByte/ags-extend-sdk-mcp-server gives an AI assistant knowledge of the Extend SDK for Go, C#, Java, and Python.