Integrations overview
Connect an AI client to Convoy, install the task Skill, or automate work from a terminal.
Use integrations to read and update Convoy work from your existing tools.
You do not need a running convoy connect process for any of these paths.
That process is for executing AI threads.
| What you want to do | Start here |
|---|---|
| Read or update tasks from an editor or AI client | Connect an MCP client with OAuth or an API key |
| Teach a local AI client the Convoy task workflow | Install Skills for AI clients |
| Automate task updates from a shell or CI | Scripting with one-shot commands |
MCP authentication and tool permissions belong to the signed-in user. Use a project-scoped endpoint to keep the client's work in the intended project. For scripts and machines without a browser, use an API key.
Local client Skills are separate from Skills in Convoy, which you manage in the web app and attach to project work.
Next steps
- MCP workflows and scoping explains project selection and repository configuration.
- MCP troubleshooting covers connection and authentication failures.
- MCP tools and CLI commands list the available operations.