GitHub MCP server | Atomicwork

Connect the GitHub MCP server

Integrate the GitHub MCP server with Atomicwork to automate repositories, issues, and workflows.

Integrate the GitHub MCP (Model Context Protocol) server with Atomicwork to give your AI Coworkers and automated workflows access to a comprehensive suite of repository, organization, and issue management tools. By connecting this server, your AI Workforce can securely interact with your GitHub environment to automate complex operations, manage code repositories, track issues, and collaborate on development tasks directly within your workspace.

Meet the prerequisites

To connect the GitHub MCP server, you must have the following permissions:

Atomicwork admin access: You need organization admin access in Atomicwork.
GitHub admin access: You need administrator access in GitHub to authorize the OAuth flow and grant the necessary permissions.

Connect the GitHub MCP server

Follow these steps to set up the integration:

Go to AI Workforce > AI Registry > MCP store > GitHub in your Atomicwork account.
Click the GitHub tile and select Connect.
Complete the OAuth authorization flow in the window that appears by logging in with your GitHub administrator credentials and authorizing the connection.
View the list of available GitHub tools displayed on your screen once the setup is complete.

Category	Permissions	Read/Write access required
Repository	Administration	Manage repo-level settings: collaborators & their access, branches and branch protection rules, merge settings, repo visibility, transfer/archive/delete. Read = view settings; Write = change them.
	Issues	Issues plus their comments, assignees, labels, and milestones. Read = view; Write = create/edit/close/comment. (Labels & milestones are shared with Pull requests.)
	Pull requests	PRs plus comments, assignees, labels, milestones, requested reviewers, and merges. Read = view; Write = open/edit/merge/comment.
Organization	API Insights	View the org’s API request analytics (usage/rate-limit insights by actor). Read-only.
	Administration	View org-level settings (org profile, security/policy config). Write would allow changing them.
	Custom organization roles	View the org’s custom repository/org roles and their definitions. Write would allow creating/editing roles.
	Members	Org membership, teams, and team membership. Write = invite/remove members, create/edit teams, manage team membership.
	Organization codespaces	Manage Codespaces owned by the org (list, stop, delete) and related secrets/settings depending on level.
	Personal access tokens	List pending fine-grained PAT requests from members. Write is needed to actually approve/deny them.
Account	Email addresses	Read the authenticated user’s email addresses (primary, verified, visibility).
	Followers	Read who the user follows and who follows them.
	Profile	Read and write to the user’s own profile — name, bio, location, company, etc.

Configure AI Coworker and builder access

After completing the setup, you can control which tools your AI Coworkers can use. This allows you to practice the principle of least privilege by granting only the necessary capabilities to each agent.

Go to AI Workforce > AI Coworker > Tools.
Select the specific AI Coworker you want to configure.
Grant granular or complete access to the available GitHub tools based on your requirements.

In addition to standard agents, the AI Assist workflow builder and the coding agent automatically have access to the MCP server while building workflows and writing custom code scripts respectively.

Understand available tool categories

The available tools list depends on your software platform configuration and can be updated as needed. The broad tool categories include:

Repository management: Create, update, delete, fork, and star repositories, manage repository collaborators, and handle repository branch protection or rulesets.
Issue and project management: Create, update, and delete issues, add assignees or labels, manage sub-issues, and configure project fields or items.
Pull requests and workflows: Create, update, and merge pull requests, review deployment protection rules, and trigger or cancel workflow runs.
Codespaces and secrets: Manage codespaces for users, configure environment or organization secrets, and control access restrictions.

Explore next steps

Now that you have connected the GitHub MCP server, see our guide on Setting up an AI Coworker to learn how to enable these tools for your digital teammates. You can also explore Code actions to see how the coding agent can utilize connected integrations in custom scripts.