opensource/coder
3
0
Fork 0
mirror of https://github.com/coder/coder.git synced 2026-09-24 15:04:27 +08:00
Code Issues Packages Projects Releases Wiki Activity

feat: allow overriding advisor reasoning effort (#27196)

Browse Source
This commit is contained in:
Michael Suchacz
2026-07-14 11:01:51 +02:00
committed by GitHub
parent 162bd0e015
commit 55b06f14a3
11 changed files with 335 additions and 78 deletions
Download Patch File Download Diff File Expand all files Collapse all files
coderd/exp_chats.go
+19 -4
View File
@@ -5572,6 +5572,9 @@ func (api *API) getChatAdvisorConfig(rw http.ResponseWriter, r *http.Request) {
}
resp.MaxUsesPerRun = max(resp.MaxUsesPerRun, 0)
resp.MaxOutputTokens = max(resp.MaxOutputTokens, 0)
if resp.ModelConfigID == uuid.Nil {
resp.ReasoningEffort = nil
}
resp.Enabled = api.Experiments.Enabled(codersdk.ExperimentChatAdvisor)
httpapi.Write(ctx, rw, http.StatusOK, resp)
@@ -5601,13 +5604,21 @@ func (api *API) putChatAdvisorConfig(rw http.ResponseWriter, r *http.Request) {
})
return
}
if req.ModelConfigID != uuid.Nil {
if req.ModelConfigID == uuid.Nil {
if req.ReasoningEffort != nil {
httpapi.Write(ctx, rw, http.StatusBadRequest, codersdk.Response{
Message: "reasoning_effort requires model_config_id.",
})
return
}
} else {
// Use system context because GetChatModelConfigByID requires
// deployment-config read access, which can be broader than the
// handler's explicit update check. The lookup only validates that
// the referenced model exists before persisting deployment config.
// handler's explicit update check. The lookup validates the model and
// any selected reasoning effort before persisting deployment config.
//nolint:gocritic // This admin-authorized validation lookup intentionally bypasses read authz.
if _, err := api.Database.GetChatModelConfigByID(dbauthz.AsSystemRestricted(ctx), req.ModelConfigID); err != nil {
modelConfig, err := api.Database.GetChatModelConfigByID(dbauthz.AsSystemRestricted(ctx), req.ModelConfigID)
if err != nil {
if errors.Is(err, sql.ErrNoRows) || httpapi.Is404Error(err) {
httpapi.Write(ctx, rw, http.StatusBadRequest, codersdk.Response{
Message: fmt.Sprintf("model_config_id %q does not match any existing model config.", req.ModelConfigID),
@@ -5620,6 +5631,10 @@ func (api *API) putChatAdvisorConfig(rw http.ResponseWriter, r *http.Request) {
})
return
}
if status, response := validateChatModelOverrideEffort(modelConfig, req.ReasoningEffort); response != nil {
httpapi.Write(ctx, rw, status, *response)
return
}
}
raw, err := json.Marshal(req)
coderd/exp_chats_test.go
+56 -2
View File
@@ -13988,13 +13988,21 @@ func TestChatAdvisorConfig_RoundTripModelConfigID(t *testing.T) {
adminClient := newChatClient(t)
coderdtest.CreateFirstUser(t, adminClient.Client)
modelConfig := createChatModelConfig(t, adminClient)
modelConfig := createAdditionalChatModelConfigWithReasoningEffort(
t,
adminClient,
"openai",
"gpt-5.2",
codersdk.ChatModelReasoningEffortMedium,
codersdk.ChatModelReasoningEffortXHigh,
)
want := codersdk.AdvisorConfig{
Enabled: true,
MaxUsesPerRun: 3,
MaxOutputTokens: 2048,
ModelConfigID: modelConfig.ID,
ReasoningEffort: ptr.Ref(codersdk.ChatModelReasoningEffortHigh),
}
err := adminClient.UpdateChatAdvisorConfig(ctx, want)
@@ -14021,6 +14029,44 @@ func TestChatAdvisorConfig_InvalidModelConfigID(t *testing.T) {
require.Contains(t, sdkErr.Message, "does not match any existing model config")
}
func TestChatAdvisorConfig_ReasoningEffortRequiresModelConfig(t *testing.T) {
t.Parallel()
ctx := testutil.Context(t, testutil.WaitLong)
adminClient := newChatClient(t)
coderdtest.CreateFirstUser(t, adminClient.Client)
err := adminClient.UpdateChatAdvisorConfig(ctx, codersdk.UpdateAdvisorConfigRequest{
ReasoningEffort: ptr.Ref(codersdk.ChatModelReasoningEffortHigh),
})
sdkErr := requireSDKError(t, err, http.StatusBadRequest)
require.Equal(t, "reasoning_effort requires model_config_id.", sdkErr.Message)
}
func TestChatAdvisorConfig_ReasoningEffortMustBeSelectable(t *testing.T) {
t.Parallel()
ctx := testutil.Context(t, testutil.WaitLong)
adminClient := newChatClient(t)
coderdtest.CreateFirstUser(t, adminClient.Client)
modelConfig := createAdditionalChatModelConfigWithReasoningEffort(
t,
adminClient,
"openai",
"gpt-5.2",
codersdk.ChatModelReasoningEffortLow,
codersdk.ChatModelReasoningEffortMedium,
)
err := adminClient.UpdateChatAdvisorConfig(ctx, codersdk.UpdateAdvisorConfigRequest{
ModelConfigID: modelConfig.ID,
ReasoningEffort: ptr.Ref(codersdk.ChatModelReasoningEffortHigh),
})
sdkErr := requireSDKError(t, err, http.StatusBadRequest)
require.Equal(t, "Invalid reasoning_effort value.", sdkErr.Message)
require.Equal(t, "Must be one of none, minimal, low, medium.", sdkErr.Detail)
}
func TestChatAdvisorConfig_RoundTripZeroValues(t *testing.T) {
t.Parallel()
@@ -14053,13 +14099,21 @@ func TestChatAdvisorConfig_OverwriteClearsPreviousValues(t *testing.T) {
adminClient := newChatClient(t)
coderdtest.CreateFirstUser(t, adminClient.Client)
modelConfig := createChatModelConfig(t, adminClient)
modelConfig := createAdditionalChatModelConfigWithReasoningEffort(
t,
adminClient,
"openai",
"gpt-5.2",
codersdk.ChatModelReasoningEffortMedium,
codersdk.ChatModelReasoningEffortXHigh,
)
rich := codersdk.AdvisorConfig{
Enabled: true,
MaxUsesPerRun: 5,
MaxOutputTokens: 1024,
ModelConfigID: modelConfig.ID,
ReasoningEffort: ptr.Ref(codersdk.ChatModelReasoningEffortHigh),
}
err := adminClient.UpdateChatAdvisorConfig(ctx, rich)
require.NoError(t, err)
coderd/x/chatd/advisor_internal_test.go
+11 -1
View File
@@ -316,6 +316,10 @@ func TestResolveAdvisorModelOverride(t *testing.T) {
providerID := uuid.New()
rawOptions, err := json.Marshal(codersdk.ChatModelCallConfig{
Temperature: func() *float64 { v := 0.42; return &v }(),
ReasoningEffort: &codersdk.ChatModelReasoningEffortConfig{
Default: ptr.Ref(codersdk.ChatModelReasoningEffortLow),
Max: ptr.Ref(codersdk.ChatModelReasoningEffortXHigh),
},
})
require.NoError(t, err)
store := &advisorOverrideStubStore{
@@ -343,7 +347,10 @@ func TestResolveAdvisorModelOverride(t *testing.T) {
gotModel, gotCfg := p.resolveAdvisorModelOverrideOrFallback(
ctx,
database.Chat{},
codersdk.AdvisorConfig{ModelConfigID: configID},
codersdk.AdvisorConfig{
ModelConfigID: configID,
ReasoningEffort: ptr.Ref(codersdk.ChatModelReasoningEffortHigh),
},
fallbackModel,
fallbackCallConfig,
modelBuildOptions{ActiveAPIKeyID: uuid.NewString()},
@@ -359,6 +366,9 @@ func TestResolveAdvisorModelOverride(t *testing.T) {
require.Equal(t, "gpt-5.2", gotModel.Model())
require.NotNil(t, gotCfg.Temperature)
require.InDelta(t, 0.42, *gotCfg.Temperature, 1e-9)
require.NotNil(t, gotCfg.ReasoningEffort)
require.Equal(t, codersdk.ChatModelReasoningEffortHigh, *gotCfg.ReasoningEffort.Default)
require.Equal(t, codersdk.ChatModelReasoningEffortHigh, *gotCfg.ReasoningEffort.Max)
})
t.Run("AIProviderIDResolvesOverrideProviderKeys", func(t *testing.T) {
t.Parallel()
coderd/x/chatd/chatd.go
+15 -2
View File
@@ -350,6 +350,19 @@ func (p *Server) resolveAdvisorModelOverride(
return fallbackModel, fallbackCallConfig, nil
}
if advisorCfg.ReasoningEffort != nil {
resolvedEffort := chatprovider.ResolveReasoningEffort(
advisorCfg.ReasoningEffort,
overrideCallConfig.ReasoningEffort,
)
if resolvedEffort != nil {
overrideCallConfig.ReasoningEffort = &codersdk.ChatModelReasoningEffortConfig{
Default: resolvedEffort,
Max: resolvedEffort,
}
}
}
return overrideModel, overrideCallConfig, nil
}
@@ -398,8 +411,8 @@ func (p *Server) newAdvisorRuntime(
}
advisorCallConfig.MaxOutputTokens = ptr.Ref(maxOutputTokens)
// The advisor has no per-turn effort selection; its model config's
// default effort applies.
// The override resolver pins an explicit advisor effort into the model
// config. Fallback models keep their configured default effort.
advisorReasoningEffort := chatprovider.ResolveReasoningEffort(
nil,
advisorCallConfig.ReasoningEffort,