Getting started
What is Toolport?
Understand the client, server and repository boundaries before you connect.
Toolport gives an MCP-compatible agent four tools for reading issue-tracker context. The client translates your request into a named tool call; Toolport reads one authorized repository and returns structured data. It does not create issues, edit fields or post comments.
The three parts of a request
Client: the application where you ask a question and approve tool use.
Server: Toolport, which advertises input schemas and reads the tracker.
Repository: the owner/repository you name in every call, such as acme/sdk.
MCP is the interface between the client and server. The tracker credential controls which repositories the server may read. Installing a server entry does not grant repository access, and repository access does not automatically enable tools inside a client.
Choose the right tool
Task | Tool |
|---|---|
Find matching issues | search_issues |
Read a known issue in full | get_issue |
Discover exact label names | list_labels |
Discover milestone titles and states | list_milestones |
What a result means
A search result contains issue summaries, not the whole conversation. Use get_issue for the description. Comments, attachments and revision history are outside this contract. A successful empty list means the call completed but nothing matched; it is different from an authorization failure.
Start here
Follow Quickstart, then make the First request. Use the tool references when you need exact field names. These pages document the Toolport v0.4 example contract; the template does not include the server implementation.