Testing code that calls an LLM: yes, you actually can
You cannot test the model, but you can test your code: snapshot the prompt as a golden file and mock the response to cover the ugly cases.
Tags
73 pages
Go
The AI provider that isn't an API
Five AI providers behind one Go interface, including one that shells out to a locally authenticated CLI for networks that allow no API calls.
AI conversations you can resume
Persisting an AI conversation in Go: snapshot the messages, prompt and tool metadata, and deliberately never the handlers or the API token.
An AI agent that has to make the build pass
Generate-and-hope does not work for codegen. Handing the draft to a repair agent with a fixed toolset that has to make build, test and lint pass.
Stop regex-ing the LLM's prose
Stop parsing an LLM's prose with regular expressions. Derive a JSON schema from your Go or Rust type so the schema and the type cannot drift.
Telemetry that asks first
Opt-in telemetry for a command-line tool: off until someone says yes, no personal data, a pluggable backend, and a consent flow that meets GDPR.
Letting the AI call your Go functions
Tool calling and the reason-act-observe loop explained, and how to define a tool in Go so the schema comes from the type rather than by hand.
Nobody reads the manual
Documentation loses to --help on location, so embed the docs in the binary: a terminal browser and an assistant grounded only in your own pages.
BDD where it earns its place, and nowhere else
BDD earns its place on narratives such as a lifecycle state machine, and nowhere else. Keep table-driven Go tests as the baseline everywhere.
An AI interface that fits on one screen
You do not need LangChain in a command-line tool. A four-method chat interface can hide five providers, tool calling and structured output.
