Every agent's model. One place. Codex on DeepSeek, Claude Code on Kimi, from the menu bar.
package gateway
import (
"encoding/json"
"net/http"
"net/http/httptest"
"os"
"path/filepath"
"strings"
"testing"
"time"
"github.com/yetone/magpie/internal/plugin"
"github.com/yetone/magpie/internal/provider"
)
// A ChatGPT backend slow to hand over its model list doesn't hold magpie's
// answer past the 5 s Codex waits for it: Codex then kept the list it was
// built with, none of magpie's models in it (#539). The answer comes in
// time, with magpie's models and Codex's own from its last list.
func TestCodexModelListBackendSlow(t *testing.T) {
setup(t, provider.Chat, &fake{t: t})
release := make(chan struct{})
up := chatgpt(t, func(w http.ResponseWriter, r *http.Request) {
select {
case <-release:
case <-r.Context().Done():
}
})
t.Cleanup(func() { close(release); up.CloseClientConnections() })
home, _ := os.UserHomeDir()
os.MkdirAll(filepath.Join(home, ".codex"), 0o755)
os.WriteFile(filepath.Join(home, ".codex", "models_cache.json"),
[]byte(`{"etag":"\"v1\"","models":[{"slug":"gpt-5.5","priority":1,"base_instructions":"x"}]}`), 0o644)
was := codexModelsWait
codexModelsWait = 200 * time.Millisecond
t.Cleanup(func() { codexModelsWait = was })
rec := httptest.NewRecorder()
req := httptest.NewRequest("GET", CodexPath+"/models?client_version=0.159.2", nil)
req.Header.Set("Authorization", "Bearer chatgpt-token")
done := make(chan struct{})
go func() { New().Handler().ServeHTTP(rec, req); close(done) }()
select {
case <-done:
case <-time.After(3 * time.Second):
t.Fatal("magpie's model list waited on the backend past the time Codex gives it")
}
var list struct {
Models []map[string]any `json:"models"`
}
json.Unmarshal(rec.Body.Bytes(), &list)
has := map[string]bool{}
for _, m := range list.Models {
s, _ := m["slug"].(string)
has[s] = true
}
if rec.Code != 200 || !has["fake/m1"] || !has["gpt-5.5"] {
t.Fatalf("%d %s", rec.Code, rec.Body.String())
}
}
// One /models request from Codex builds magpie's providers once (#746,
// sperwe: seven builds, each reading every agent's sign-in, some 200
// keychain reads, took 5.3–5.6 s, past the 5 s Codex waits, and Codex kept
// its own list): the catalog, the codex provider's picks and windows and
// the list's tag all read the one build.
func TestCodexModelListBuildsProvidersOnce(t *testing.T) {
// not from the plugins' list an earlier test's folder had: seen last,
// it reads as changed in setup's folder, and the change, told in the
// background, has the providers built again for the list
plugin.UseCached(nil)
setup(t, provider.Chat, &fake{t: t})
chatgpt(t, func(w http.ResponseWriter, r *http.Request) {
w.Write([]byte(`{"models":[{"slug":"gpt-5.5","priority":1}]}`))
})
h := New().Handler()
n := provider.AllBuilt.Load()
rec := httptest.NewRecorder()
req := httptest.NewRequest("GET", CodexPath+"/models?client_version=0.159.2", nil)
req.Header.Set("Authorization", "Bearer chatgpt-token")
h.ServeHTTP(rec, req)
if rec.Code != 200 || !strings.Contains(rec.Body.String(), `"fake/m1"`) || !strings.Contains(rec.Body.String(), `"gpt-5.5"`) {
t.Fatalf("%d %s", rec.Code, rec.Body.String())
}
if got := provider.AllBuilt.Load() - n; got != 1 {
t.Fatalf("the providers were built %d times for one list, want 1", got)
}
}