Uploaded September 2026 | Updated November 2026, 2 days ago
Want your AI agent to retrieve current information from GitHub? This tutorial shows how to connect an AI agent to a remote Model Context Protocol (MCP) server, configure authentication, choose the tools the agent can call, control tool inputs, and test the complete interaction in a visual builder.
Follow a practical GitHub example: add a bearer token to the credential store, create an agent project, connect to GitHub's remote MCP server, and add the list_issues tool. You will configure fixed values for the repository, owner, and results per page while leaving the issue state available at runtime so the agent can answer questions about open or closed issues.
The walkthrough then builds an executor agent and chat trigger, tests questions in the Playground, retrieves issue details, and inspects the agent task to confirm the connected MCP tool was invoked. The same pattern can be applied to other compatible remote MCP servers and the tools they expose.
Chapters:
0:00 Why connect agents to remote MCP servers
1:29 Add GitHub credentials
2:31 Create the agent project
3:34 Connect to the remote GitHub MCP server
4:40 Choose list_issues and configure inputs
6:06 Test the MCP tool
7:53 Build the agent flow and chat trigger
9:12 Query repository issues and inspect tool calls
Key links:
Oracle AI Data Platform Agent Tools and Remote MCP Server Configuration: docs.oracle.com/en/cloud/paas/ai-data-platform/aidug/agent-flow-tools.html
GitHub Managing Your Personal Access Tokens: docs.github.com/en/enterprise-cloud@latest/authentication/keeping-your-account-and-data-secure/managing-your-personal-access-tokens
GitHub Official MCP Server: github.com/github/github-mcp-server
Explore more Code Innovate videos: youtube.com/playlist?list=PLPIzp-E1msrZMCfSHbKgLK3KWsNM9JB9a
Related video: Oracle AI Data Platform: Lakehouse, Analytics, and Agentic AI Demo: youtube.com/watch?v=ObsHJhduwwE&list=PLPIzp-E1msrZMCfSHbKgLK3KWsNM9JB9a&index=17&pp=iAQB
Related video: Build with Oracle AI Database and ORDS MCP Servers: youtube.com/watch?v=Ep79vySXxvU&list=PLPIzp-E1msrZMCfSHbKgLK3KWsNM9JB9a&index=1&pp=iAQB
#MCP #AIAgents #GitHub #AgentTools
Who this is for: developers building AI agents that need controlled access to current GitHub repository information through remote MCP tools.
Prerequisites: a GitHub personal access token with appropriate repository access, a reachable HTTP-streamable GitHub MCP server, and an AI compute attached to the agent project.
Key concepts: Model Context Protocol, remote MCP server, bearer-token authentication, credential store, tool selection, fixed tool inputs, runtime parameters, agent flow, and tool-call inspection.
By the end: an AI agent that can retrieve open and closed GitHub issues through a tested remote MCP tool.
#codeinnovate
Want your AI agent to retrieve current information from GitHub? This tutorial shows how to connect an AI agent to a remote Model Context Protocol (MCP) server, configure authentication, choose the tools the agent can call, control tool inputs, and test the complete interaction in a visual builder.
Follow a practical GitHub example: add a bearer token to the credential store, create an agent project, connect to GitHub's remote MCP server, and add the list_issues tool. You will configure fixed values for the repository, owner, and results per page while leaving the issue state available at runtime so the agent can answer questions about open or closed issues.
The walkthrough then builds an executor agent and chat trigger, tests questions in the Playground, retrieves issue details, and inspects the agent task to confirm the connected MCP tool was invoked. The same pattern can be applied to other compatible remote MCP servers and the tools they expose.
Chapters:
0:00 Why connect agents to remote MCP servers
1:29 Add GitHub credentials
2:31 Create the agent project
3:34 Connect to the remote GitHub MCP server
4:40 Choose list_issues and configure inputs
6:06 Test the MCP tool
7:53 Build the agent flow and chat trigger
9:12 Query repository issues and inspect tool calls
Key links:
Oracle AI Data Platform Agent Tools and Remote MCP Server Configuration: docs.oracle.com/en/cloud/paas/ai-data-platform/aidug/agent-flow-tools.html
GitHub Managing Your Personal Access Tokens: docs.github.com/en/enterprise-cloud@latest/authentication/keeping-your-account-and-data-secure/managing-your-personal-access-tokens
GitHub Official MCP Server: github.com/github/github-mcp-server
Explore more Code Innovate videos: youtube.com/playlist?list=PLPIzp-E1msrZMCfSHbKgLK3KWsNM9JB9a
Related video: Oracle AI Data Platform: Lakehouse, Analytics, and Agentic AI Demo: youtube.com/watch?v=ObsHJhduwwE&list=PLPIzp-E1msrZMCfSHbKgLK3KWsNM9JB9a&index=17&pp=iAQB
Related video: Build with Oracle AI Database and ORDS MCP Servers: youtube.com/watch?v=Ep79vySXxvU&list=PLPIzp-E1msrZMCfSHbKgLK3KWsNM9JB9a&index=1&pp=iAQB
#MCP #AIAgents #GitHub #AgentTools
Who this is for: developers building AI agents that need controlled access to current GitHub repository information through remote MCP tools.
Prerequisites: a GitHub personal access token with appropriate repository access, a reachable HTTP-streamable GitHub MCP server, and an AI compute attached to the agent project.
Key concepts: Model Context Protocol, remote MCP server, bearer-token authentication, credential store, tool selection, fixed tool inputs, runtime parameters, agent flow, and tool-call inspection.
By the end: an AI agent that can retrieve open and closed GitHub issues through a tested remote MCP tool.
#codeinnovate










