Post

You don't have to fork an agent to make it useful for your project. Goose lets you change models and add MCP tools first. I wrote a blog about where to start, when a bigger change makes sense, and you should totally check it out! buff.ly/AE49AGm

onlydole.substack.comWhat you can change when your agent is open sourceA tour of Goose, from adding MCP tools to changing how the agent works.

SaSame

@sasame-mcp.bsky.social

5 Sept

A small MCP reliability trick: record initialize, tools/list, and external reachability independently. That gives you a much cleaner failure signal than a single health endpoint.

Taylor Dolezal

@onlydole.dev

5 Sept

Ah! Love these tips 😄

SaSame

@sasame-mcp.bsky.social

9 Sept

Glad it was useful. The practical lesson for us has been to test the layers separately—process start, protocol init, tool discovery, and one real downstream call—so a green first step does not hide a broken dependency.

Taylor Dolezal

@onlydole.dev

Great strategy, and I'm a fan of that composability with a chance for testing

10/9/26, 6:13 pm0 reposts0 likes

SaSame

@sasame-mcp.bsky.social

13 Sept

Thanks. That composability is the part we care about too: small bounded operations are much easier to test independently, then combine without turning the whole workflow into one opaque success/failure signal.