Back to timeline

Thu, October 1509:15ToolsMCP & SkillsMCP & Skills guide

mcp-go implements MCP spec 2025-11-25 with task-enhanced tools

Decision Brief

What changedThe GitHub project mark3labs/mcp-go, a Go implementation of the Model Context Protocol server library, now supports the 2025-11-25 spec and adds asynchronous task-enhanced tools.
Why it mattersGo developers building MCP servers can now use AddTaskTool and WithTaskCapabilities to turn long-running tools into async tasks that clients poll for results, while WithMaxConcurrentTasks caps server-side concurrency to avoid blocking or timeouts.
Who should careAI coding tool users, Agent builders
Affected stackMCP
Builder actionCheck whether your existing MCP servers are affected
Source confidenceHigh · Official release / blog / repo

mcp-go is a Go implementation of the Model Context Protocol, used to connect LLM applications with external data sources and tools. It implements MCP spec 2025-11-25 and remains backward compatible with 2025-06-18, 2025-03-26, and 2024-11-05. Core capabilities include servers, resources, tools, and prompts; resources support static URIs and dynamic templates, and tools can run either synchronously or as task-enhanced async operations. Task-enhanced tools offer three modes: TaskSupportForbidden, TaskSupportOptional, and TaskSupportRequired. In async mode the server immediately returns a task ID, the tool runs in the background, clients poll results via tasks/result, and the server sends status notifications on completion. For Go developers writing MCP servers, async task mode keeps long-running operations from blocking or timing out; WithMaxConcurrentTasks limits the number of tasks in the working state, with new task creation requests failing once the cap is reached. Completed, failed, or canceled tasks don't count toward the limit, and unset or zero means no limit. Additional features such as OAuth-protected resource metadata, session management, per-session tools, tool filtering, request hooks, and tool-handling middleware give teams building full MCP services infrastructure from authentication to tool governance. The project is still in active development: core features are usable, while some advanced capabilities are still in progress.

Summary basis: full article readCompiled from the source scope noted above; the original remains authoritative.

Sources

  • Skill Radar(GitHub 趋势)

    Trending hands-on MCP servers, agent skills, and AI-coding tools discovered from GitHub search momentum.

  • GitHub:mark3labs/mcp-go

Related intel

留言

登入后即可留言,和其他 builder 交换实测心得。

还没有留言,抢头香。