name	go-test
description	Run go test with race detector, analyze failures, fix issues, and re-run until green
license	MIT
metadata	{"audience":"developers","workflow":"testing"}

Run the Go test suite with the race detector, analyze any failures, apply fixes, and iterate until all tests pass.

Use this skill before committing changes, before opening a PR, or when asked to fix failing tests.

Run the full test suite:
```
go test ./... -race -v
```
Analyze failures — for each failing test:
- Identify the test function and file (e.g., gateway/methods_test.go:TestChatSend)
- Read the test code and the code under test
- Determine root cause: type mismatch, missing mock, logic error, race condition, etc.
Fix the issue — apply the minimal correct fix:
- If the test expectation is wrong (API changed), update the test
- If the implementation is wrong, fix the implementation
- If a new type/method is missing, add it
- Never delete or skip tests to make the suite pass
Re-run tests:
```
go test ./... -race
```
Repeat steps 2-4 until all tests pass.
Run vet as a final check:
```
go vet ./...
```

Table-driven tests — most tests use []struct{ name string; ... } with t.Run()
Mock gateway — gateway/methods_test.go uses a mock WebSocket server that validates request frames and sends response frames
Race detector — always use -race; the gateway client is concurrent and race bugs are real
Coverage target — aim for 100% statement coverage on library packages

File	What it covers
`gateway/methods_test.go`	All 96+ typed RPC methods
`gateway/client_test.go`	Connection lifecycle, handshake, reconnect
`protocol/types_test.go`	Serialization round-trips
`protocol/parse_test.go`	Frame parsing edge cases
`chatcompletions/client_test.go`	HTTP client, SSE streaming
`openresponses/client_test.go`	Responses API, SSE events
`toolsinvoke/client_test.go`	Tools invoke HTTP client
`discovery/browser_test.go`	mDNS discovery
`acp/server_test.go`	JSON-RPC dispatch, handler interface