Concord MCP. Manage CI/CD status and logs from chat.

Look, you're tired of jumping between a dozen dashboards just to see if your code deployed right. Concord lets your AI agent run complex CI/CD pipelines and workflow orchestrations straight from chat. You don't gotta leave your terminal; it handles the whole mess so you can keep talking to your agent while things happen in the background.

It’s all about control. Your agent talks directly to Concord, treating massive pipeline operations like simple conversation turns. When you need to get started, you first list everything up. You can use list_organizations to pull a full roster of every top-level organizational container set up in your instance. Then, if you know which department it is, you run list_projects to see all the individual projects hanging off that organization.

From there, you grab list_repositories, and the agent pulls a list of every source code repo connected and ready for work on that project.

Once everything's set up, keeping an eye on what’s running is simple. You can check if anything’s live right now by using list_running_processes; it gives you an immediate status report on every process currently executing across all projects. If you need details on a specific job—like when it started or how long it's been chewing up CPU cycles—you run get_process_details to pull that detailed metadata for a single execution instance.

When the smoke clears, and you’re done monitoring live runs, you can check out historical data. You use list_processes to generate a list of every completed or failed workflow process execution across all scopes. If you find an ID that looks sketchy, you can run get_process_log, which fetches the complete, raw log output for any given process ID; this is your go-to tool for diagnosing failures because it gives you everything.

And when things get sticky, you take action. You trigger a brand new workflow whenever necessary by running start_process for a specified project or template. If that deployment goes sideways and you need to kill it—maybe it's eating bandwidth or doing something wrong—you use terminate_process to stop the running process execution immediately.

This is crucial for rollbacks or just general cleanup.

These tools let your agent manage everything programmatically. You initiate new runs, check status, pull logs, and kill processes all from chat. It means you audit project structures, monitor deployments, and resolve incidents without ever having to switch context or leave the conversation flow.