feat(dashboard): redesign, mobile UI, runtime provider/fallback/pool management

.gitignore

@@ -0,0 +1,3 @@
+node_modules/
+pnpm-lock.yaml
+резюме/

configs/config.example.yaml

@@ -325,6 +325,30 @@ token_saver:
   audit:
     enabled: true # env: TOKEN_SAVER_AUDIT_ENABLED
 
+# ========= RESPONSE HEADERS =========
+# Additional debug headers added to gateway responses
+response_headers:
+  # Enable adding X-Actual-Provider / X-Actual-Model / X-Requested-Model / X-Fallback-Chain headers
+  enabled: false # env: RESPONSE_HEADERS_ENABLED
+  # Add headers when a fallback was used during request execution
+  include_fallback: true # env: RESPONSE_HEADERS_INCLUDE_FALLBACK
+  # Add headers when no fallback was used (normal request execution)
+  include_non_fallback: true # env: RESPONSE_HEADERS_INCLUDE_NON_FALLBACK
+  # Emit the X-Actual-Provider header (the provider that actually served the response)
+  actual_provider_header: true # env: RESPONSE_HEADERS_ACTUAL_PROVIDER
+  # Emit the X-Actual-Model header (the model that actually produced the response)
+  actual_model_header: true # env: RESPONSE_HEADERS_ACTUAL_MODEL
+  # Emit the X-Requested-Model header (the originally requested model selector)
+  requested_model_header: true # env: RESPONSE_HEADERS_REQUESTED_MODEL
+  # Emit the X-Fallback-Chain header (comma-separated models attempted before a success)
+  fallback_chain_header: true # env: RESPONSE_HEADERS_FALLBACK_CHAIN
+  # Custom response headers. Values support placeholders: {actual_provider},
+  # {actual_model}, {requested_model}, {fallback_chain}, {request_id}
+  custom_headers:
+    - name: X-Custom-Header
+      value: "{actual_model} via {actual_provider}"
+      enabled: true
+
 # ========= WORKFLOWS =========
 # Workflow execution and refresh
 workflows:

configs/dashboard-overrides.yaml

@@ -100,3 +100,15 @@ token_saver:
     on_error: allow
     emit_headers: false
     audit_enabled: false
+response_headers:
+    enabled: false
+    include_fallback: true
+    include_non_fallback: true
+    actual_provider_header: true
+    actual_model_header: true
+    requested_model_header: true
+    fallback_chain_header: true
+    custom_headers:
+        - name: X-Custom-Header
+          value: "{actual_model} via {actual_provider}"
+          enabled: true

configs/fallback.example.json

@@ -1,10 +1,18 @@
-{
-  "gpt-4o": [
-    "azure/gpt-4o",
-    "gemini/gemini-2.5-pro"
-  ],
-  "claude-sonnet-4": [
-    "openai/gpt-5-mini",
-    "gemini/gemini-2.5-pro"
-  ]
-}
+[
+  {
+    "source": "gpt-4o",
+    "targets": [
+      "azure/gpt-4o",
+      "gemini/gemini-2.5-pro"
+    ],
+    "enabled": true
+  },
+  {
+    "source": "claude-sonnet-4",
+    "targets": [
+      "openai/gpt-5-mini",
+      "gemini/gemini-2.5-pro"
+    ],
+    "enabled": true
+  }
+]

configuration/config.go

@@ -30,23 +30,24 @@ var bodySizeLimitRegex = regexp.MustCompile(`(?i)^(\d+)([KMG])?B?$`)
 
 // Config holds the application configuration.
 type Config struct {
-	Server               ServerConfig               `yaml:"server"`
-	Models               ModelsConfig               `yaml:"models"`
-	Cache                CacheConfig                `yaml:"cache"`
-	Storage              StorageConfig              `yaml:"storage"`
-	Logging              LogConfig                  `yaml:"logging"`
-	Usage                UsageConfig                `yaml:"usage"`
-	Metrics              MetricsConfig              `yaml:"metrics"`
-	HTTP                 HTTPConfig                 `yaml:"http"`
-	Admin                AdminConfig                `yaml:"admin"`
-	Guardrails           GuardrailsConfig           `yaml:"guardrails"`
-	Fallback             FallbackConfig             `yaml:"fallback"`
-	Combos               CombosConfig               `yaml:"combos"`
-	CLITools             CLIToolsConfig             `yaml:"cli_tools"`
-	Workflows            WorkflowsConfig            `yaml:"workflows"`
-	Resilience           ResilienceConfig           `yaml:"resilience"`
-	TokenSaver           TokenSaverConfig           `yaml:"token_saver"`
-	Edition              EditionConfig              `yaml:"edition"`
+	Server          ServerConfig          `yaml:"server"`
+	Models          ModelsConfig          `yaml:"models"`
+	Cache           CacheConfig           `yaml:"cache"`
+	Storage         StorageConfig         `yaml:"storage"`
+	Logging         LogConfig             `yaml:"logging"`
+	Usage           UsageConfig           `yaml:"usage"`
+	Metrics         MetricsConfig         `yaml:"metrics"`
+	HTTP            HTTPConfig            `yaml:"http"`
+	Admin           AdminConfig           `yaml:"admin"`
+	Guardrails      GuardrailsConfig      `yaml:"guardrails"`
+	Fallback        FallbackConfig        `yaml:"fallback"`
+	Combos          CombosConfig          `yaml:"combos"`
+	CLITools        CLIToolsConfig        `yaml:"cli_tools"`
+	Workflows       WorkflowsConfig       `yaml:"workflows"`
+	Resilience      ResilienceConfig      `yaml:"resilience"`
+	TokenSaver      TokenSaverConfig      `yaml:"token_saver"`
+	Edition         EditionConfig         `yaml:"edition"`
+	ResponseHeaders ResponseHeadersConfig `yaml:"response_headers"`
 }
 
 // LoadResult is returned by Load and bundles the application config with the raw
@@ -68,6 +69,14 @@ type RawProviderConfig struct {
 	APIVersion string               `yaml:"api_version"`
 	Models     []RawProviderModel   `yaml:"models"`
 	Resilience *RawResilienceConfig `yaml:"resilience"`
+	// BindIP optionally sets the local outbound IP address used when this
+	// provider's HTTP client connects to the upstream. Useful when the host
+	// has multiple public IPs and the upstream rate-limits per source IP.
+	BindIP string `yaml:"bind_ip"`
+	// PoolOnly, when true, hides this provider's models from the public model
+	// list. The provider is then only reachable through a pool that lists it as
+	// a member, keeping the /v1/models list uncluttered.
+	PoolOnly bool `yaml:"pool_only"`
 }
 
 // RawPoolConfig is the YAML-sourced provider pool definition. Pools group
@@ -222,8 +231,6 @@ type AdminConfig struct {
 	// a warning is logged and UI is forced to false.
 	// Default: true
 	UIEnabled bool `yaml:"ui_enabled" env:"ADMIN_UI_ENABLED"`
-
-
 }
 
 // GuardrailsConfig holds configuration for the request guardrails pipeline.
@@ -380,6 +387,65 @@ type HTTPProxyConfig struct {
 	CACertPEM        string `yaml:"ca_cert_pem"`
 }
 
+// ResponseHeadersConfig controls the additional response headers added to
+// gateway responses for debugging and monitoring purposes.
+type ResponseHeadersConfig struct {
+	// Enabled controls whether the additional response headers are added.
+	// Default: false.
+	Enabled bool `yaml:"enabled" env:"RESPONSE_HEADERS_ENABLED"`
+
+	// Mode controls when headers are added.
+	// Values: "success" (only 2xx), "error" (only 4xx/5xx), "always" (all responses).
+	// Default: "success".
+	Mode string `yaml:"mode" env:"RESPONSE_HEADERS_MODE"`
+
+	// IncludeFallback controls whether headers are added when a fallback
+	// was used during request execution.
+	// Default: true.
+	IncludeFallback bool `yaml:"include_fallback" env:"RESPONSE_HEADERS_INCLUDE_FALLBACK"`
+
+	// IncludeNonFallback controls whether headers are added when no fallback
+	// was used (normal request execution).
+	// Default: true.
+	IncludeNonFallback bool `yaml:"include_non_fallback" env:"RESPONSE_HEADERS_INCLUDE_NON_FALLBACK"`
+
+	// ActualProviderHeader controls whether the X-Actual-Provider header
+	// (the provider that actually served the response) is added.
+	// Default: true.
+	ActualProviderHeader bool `yaml:"actual_provider_header" env:"RESPONSE_HEADERS_ACTUAL_PROVIDER"`
+
+	// ActualModelHeader controls whether the X-Actual-Model header
+	// (the model that actually produced the response) is added.
+	// Default: true.
+	ActualModelHeader bool `yaml:"actual_model_header" env:"RESPONSE_HEADERS_ACTUAL_MODEL"`
+
+	// RequestedModelHeader controls whether the X-Requested-Model header
+	// (the originally requested model selector) is added.
+	// Default: true.
+	RequestedModelHeader bool `yaml:"requested_model_header" env:"RESPONSE_HEADERS_REQUESTED_MODEL"`
+
+	// FallbackChainHeader controls whether the X-Fallback-Chain header
+	// (comma-separated models attempted before a success) is added.
+	// Default: true.
+	FallbackChainHeader bool `yaml:"fallback_chain_header" env:"RESPONSE_HEADERS_FALLBACK_CHAIN"`
+
+	// CustomHeaders are additional headers added verbatim to responses.
+	// Values support the same placeholders as the built-in headers.
+	// Default: none.
+	CustomHeaders []CustomResponseHeaderConfig `yaml:"custom_headers"`
+}
+
+// CustomResponseHeaderConfig defines a user-provided response header.
+type CustomResponseHeaderConfig struct {
+	// Name is the HTTP header name (e.g. "X-Custom-Header").
+	Name string `yaml:"name"`
+	// Value is the header value. Supported placeholders: {actual_provider},
+	// {actual_model}, {requested_model}, {fallback_chain}, {request_id}.
+	Value string `yaml:"value"`
+	// Enabled controls whether this header is emitted.
+	Enabled bool `yaml:"enabled"`
+}
+
 // WorkflowsConfig holds runtime refresh behavior for persisted workflows.
 type WorkflowsConfig struct {
 	// RefreshInterval controls how often the in-memory workflow snapshot
@@ -603,8 +669,6 @@ type MetricsConfig struct {
 	Endpoint string `yaml:"endpoint" env:"METRICS_ENDPOINT"`
 }
 
-
-
 // RetryConfig holds resolved retry settings for an LLM client.
 // This is the canonical type shared between config and llmclient.
 type RetryConfig struct {
@@ -657,7 +721,7 @@ func buildDefaultConfig() *Config {
 			BasePath:                "/",
 			SwaggerEnabled:          false,
 			PprofEnabled:            false,
-			DisableRequestLogging: true,
+			DisableRequestLogging:   true,
 			EnablePassthroughRoutes: true,
 			EnableAnthropicIngress:  false,
 			AllowPassthroughV1Alias: true,
@@ -742,6 +806,16 @@ func buildDefaultConfig() *Config {
 		Admin:      AdminConfig{EndpointsEnabled: false, UIEnabled: false},
 		Guardrails: GuardrailsConfig{},
 		TokenSaver: defaultTokenSaverConfig(),
+		ResponseHeaders: ResponseHeadersConfig{
+			Enabled:              false,
+			Mode:                 "success",
+			IncludeFallback:      true,
+			IncludeNonFallback:   true,
+			ActualProviderHeader: true,
+			ActualModelHeader:    true,
+			RequestedModelHeader: true,
+			FallbackChainHeader:  true,
+		},
 		Edition: EditionConfig{
 			Name: EditionOSS,
 		},
@@ -841,12 +915,6 @@ func applyYAML(cfg *Config) (map[string]RawProviderConfig, map[string]RawPoolCon
 	rawProviders := make(map[string]RawProviderConfig)
 	rawPools := make(map[string]RawPoolConfig)
 
-	if data == nil {
-		return rawProviders, rawPools, nil
-	}
-
-	expanded := expandString(string(data))
-
 	// yamlTarget is a local struct that mirrors Config for YAML unmarshaling,
 	// using RawProviderConfig for providers so nullable resilience overrides are preserved.
 	type yamlTarget struct {
@@ -855,16 +923,20 @@ func applyYAML(cfg *Config) (map[string]RawProviderConfig, map[string]RawPoolCon
 		RawPools     map[string]RawPoolConfig     `yaml:"pools"`
 	}
 
-	target := yamlTarget{Config: cfg}
-	if err := yaml.Unmarshal([]byte(expanded), &target); err != nil {
-		return nil, nil, fmt.Errorf("failed to parse %s: %w", configPath, err)
-	}
+	if data != nil {
+		expanded := expandString(string(data))
 
-	if target.RawProviders != nil {
-		rawProviders = target.RawProviders
-	}
-	if target.RawPools != nil {
-		rawPools = target.RawPools
+		target := yamlTarget{Config: cfg}
+		if err := yaml.Unmarshal([]byte(expanded), &target); err != nil {
+			return nil, nil, fmt.Errorf("failed to parse %s: %w", configPath, err)
+		}
+
+		if target.RawProviders != nil {
+			rawProviders = target.RawProviders
+		}
+		if target.RawPools != nil {
+			rawPools = target.RawPools
+		}
 	}
 
 	for _, overlayPath := range dashboardOverridePaths(customPath) {
@@ -911,8 +983,6 @@ func dashboardOverridePaths(customPath string) []string {
 	}
 }
 
-
-
 func validatePublicHTTPURL(raw, field string) error {
 	u, err := url.Parse(strings.TrimSpace(raw))
 	if err != nil || u.Scheme == "" || u.Host == "" {