feat(tools): stream shell_run output live into the TUI

- Arrow-key input history (up/down with multiline awareness)
- Bang shell commands via ! prefix (requires --yolo)
- Codex-style input bar with mode indicator (R/Y/⌘)
- Command palette virtual scrolling
- Multiline input viewport fix
- Merged model pills + footer into single status line
- Removed shift+enter (wait for TUI library support)
- Removed default footer hints

README.md

@@ -8,6 +8,7 @@ model/provider switching, and local tools.
 
 - TUI-first chat experience with a light theme by default.
 - OpenAI-compatible `/v1/chat/completions` provider support.
+- OpenAI Responses API (`/v1/responses`) provider support.
 - Multiple providers from `~/.agentu/config.yaml`.
 - Session-only provider, model, and thinking-level switching.
 - Read-only project tools enabled by default.
@@ -23,6 +24,12 @@ export AGENTU_API_KEY='your-api-key'
 go run .
 ```
 
+Use dark mode when your terminal is dark:
+
+```sh
+go run . --theme dark
+```
+
 Use `--yolo` only when you want agentu to write files or run shell commands:
 
 ```sh
@@ -42,6 +49,7 @@ Use `--config <path>` to load another file.
 ```yaml
 providers:
   loveuer:
+    api_type: chat
     base_url: https://ai.loveuer.com
     api_key: ${AGENTU_API_KEY}
     models:
@@ -58,6 +66,20 @@ Provider names are the keys under `providers`. At startup, agentu selects the
 first provider by name. Use `/model provider <name>` to switch providers for the
 current session.
 
+`api_type` selects which API protocol the provider speaks:
+
+```text
+chat      OpenAI-compatible /v1/chat/completions (default)
+responses OpenAI Responses API /v1/responses
+```
+
+The Responses API client converts the conversation history to Responses input
+items (`user`/`assistant` messages, `function_call`, and
+`function_call_output`), sends the system prompt as the top-level
+`instructions` field, streams `response.output_text.delta` events, and supports
+function calling through `response.output_item.added`,
+`response.function_call_arguments.delta`, and `response.output_item.done`.
+
 `models` is required. `/model model <name>` is restricted to that list. Each
 entry is an object with:
 
@@ -77,7 +99,11 @@ none, middle, high, xhigh, max
 ```
 
 When configured or changed with `/model thinking ...`, the value is sent as a
-top-level OpenAI-compatible request field named `thinking`.
+top-level request field. The default field depends on `api_type`: `chat` uses
+`thinking`, while `responses` uses the official `reasoning` field, sent as
+`{"effort": ...}` with agentu's `middle` level mapped to the official `medium`
+value (`none`, `high`, `xhigh`, `max` pass through). Override with
+`thinking_param` if a provider expects a different field name.
 
 ## CLI Flags
 
@@ -94,22 +120,25 @@ go run . [flags]
 ## TUI Controls
 
 - `enter` sends the message.
-- `ctrl+j` or `alt+enter` inserts a newline.
+- `ctrl+j` inserts a newline.
+- `up` recalls the previous input; `ctrl+p` / `ctrl+n` navigate input history.
 - The input starts at one line and grows up to six lines.
 - `page up` / `page down` scrolls the transcript.
 - `esc` cancels a running response.
 - Type `/` to show available commands.
+- In `--yolo` mode, type `! <command>` to run a shell command from the configured working directory.
 
 Slash commands:
 
 - `/clear` resets conversation history.
-- `/compact` summarizes older conversation history and keeps recent context.
+- `/compact` summarizes older conversation history and keeps the newest ~20% of context, with a small recent-message floor.
 - `/status` shows changed files with `git status --short`.
 - `/diff [--stat] [path...]` shows the unstaged git diff.
 - `/test [command...]` runs a project test command; defaults to `go test ./...` for Go modules or `npm test` for Node projects.
 - `/model` shows current provider, model, thinking, and switch usage.
 - `/model provider <name>` switches provider for the current session.
 - `/model model <name>` or `/model <name>` switches model for the current session.
+  In the TUI, type `/model ` or `/model model ` to choose from configured model IDs in the candidate palette.
 - `/model thinking <none|middle|high|xhigh|max>` changes thinking for the current session.
 - `/sessions` lists saved sessions.
 - `/resume [id]` resumes by ID or opens an interactive picker.
@@ -138,7 +167,8 @@ Web tools are also available in read-only mode:
 
 - `file_write` for creating or replacing whole files
 - `file_edit` for surgical line-range or pattern edits to existing files
-- `shell_run` for non-interactive shell commands
+- `shell_run` for non-interactive shell commands; output streams live into the
+  TUI while the command runs instead of appearing only after completion
 
 Use `--yolo` for prompts that ask agentu to run or inspect local commands, for
 example:
@@ -165,5 +195,8 @@ Local config and build outputs are intentionally ignored by git.
 
 - Provider/model/thinking changes are session-only; agentu does not write them
   back to `~/.agentu/config.yaml`.
+- `api_type` is per provider and cannot be changed at runtime; use
+  `/model provider <name>` to switch to a provider configured with a different
+  API type.
 - The config schema is intentionally strict during MVP development. Deprecated
   fields such as `provider` or `active_provider` are rejected.

agentu.example.yaml

@@ -1,5 +1,8 @@
 providers:
   loveuer:
+    # API protocol: "chat" (OpenAI-compatible /v1/chat/completions, default)
+    # or "responses" (OpenAI Responses API /v1/responses).
+    api_type: chat
     base_url: https://ai.loveuer.com
     api_key: ${AGENTU_API_KEY}
     models:

internal/agent/agent.go

@@ -15,14 +15,16 @@ import (
 )
 
 const (
-	defaultMaxToolRounds      = 50
-	doomLoopThreshold         = 3
-	compactRecentMessages     = 6
-	summaryMessageLimit       = 8000
-	contextSummaryPrefix      = "[context summary]"
-	toolLogArgumentLimit      = 400
-	toolLogOutputPreviewBytes = 4096
-	toolLogOutputMarker       = "[output]"
+	defaultMaxToolRounds          = 50
+	doomLoopThreshold             = 3
+	compactMinRecentMessages      = 6
+	compactRecentRetentionPercent = 20
+	summaryMessageLimit           = 8000
+	contextSummaryPrefix          = "[context summary]"
+	toolLogArgumentLimit          = 400
+	toolLogOutputPreviewBytes     = 4096
+	toolLogOutputMarker           = "[output]"
+	toolStreamLogLimit            = tools.MaxToolOutputBytes
 
 	summarizerSystemPrompt = "You are a conversation summarizer. Summarize the following conversation history into a concise but detailed summary. Preserve:\n" +
 		"- Key facts, decisions, and conclusions\n" +
@@ -182,6 +184,61 @@ func (a *Agent) SetLastUsage(usage *llm.Usage) {
 	a.lastUsage = &copy
 }
 
+// compactRetainedTokenBudget returns the number of estimated tokens to retain
+// based on compactRecentRetentionPercent, using integer ceiling division.
+func compactRetainedTokenBudget(totalTokens int) int {
+	if totalTokens <= 0 {
+		return 1
+	}
+	return (totalTokens*compactRecentRetentionPercent + 99) / 100
+}
+
+// compactTurnBoundary walks backward from start until it finds a user message
+// or reaches prefixEnd, ensuring the retained suffix starts at a user turn.
+func compactTurnBoundary(messages []llm.Message, prefixEnd int, start int) int {
+	for start > prefixEnd && messages[start].Role != llm.RoleUser {
+		start--
+	}
+	return start
+}
+
+// compactRecentStart returns the index where the retained recent suffix begins.
+// It balances the 20% token-budget ratio with the compactMinRecentMessages floor
+// and aligns to a user-turn boundary.
+func compactRecentStart(messages []llm.Message, prefixEnd int) int {
+	compactableCount := len(messages) - prefixEnd
+	if compactableCount <= compactMinRecentMessages {
+		return len(messages)
+	}
+
+	compactable := messages[prefixEnd:]
+	totalTokens := llm.EstimateTokens(compactable, nil)
+	retainTokens := compactRetainedTokenBudget(totalTokens)
+
+	// Walk backward from the end to find the earliest suffix fitting the budget.
+	ratioStart := len(messages) - 1
+	for i := len(messages) - 1; i >= prefixEnd; i-- {
+		if llm.EstimateTokens(messages[i:], nil) <= retainTokens {
+			ratioStart = i
+		} else {
+			break
+		}
+	}
+
+	// Enforce the minimum recent-message floor.
+	minStart := len(messages) - compactMinRecentMessages
+	if minStart < prefixEnd {
+		minStart = prefixEnd
+	}
+	// Choose the earlier start so the floor is never weakened.
+	start := minStart
+	if ratioStart < start {
+		start = ratioStart
+	}
+
+	return compactTurnBoundary(messages, prefixEnd, start)
+}
+
 // Compact summarizes older conversation messages while preserving the system
 // prompt and recent messages intact.
 func (a *Agent) Compact(ctx context.Context) error {
@@ -196,12 +253,8 @@ func (a *Agent) Compact(ctx context.Context) error {
 	if a.messages[0].Role == llm.RoleSystem {
 		prefixEnd = 1
 	}
-	if len(a.messages)-prefixEnd <= compactRecentMessages {
-		return nil
-	}
-
-	recentStart := len(a.messages) - compactRecentMessages
-	if recentStart <= prefixEnd {
+	recentStart := compactRecentStart(a.messages, prefixEnd)
+	if recentStart <= prefixEnd || recentStart >= len(a.messages) {
 		return nil
 	}
 	middle := append([]llm.Message(nil), a.messages[prefixEnd:recentStart]...)
@@ -443,32 +496,75 @@ func (a *Agent) toolDefinitions() []llm.Tool {
 	return a.toolRegistry.Definitions()
 }
 
-func (a *Agent) executeTool(ctx context.Context, call llm.ToolCall) string {
+func (a *Agent) executeTool(ctx context.Context, call llm.ToolCall, logs io.Writer) (string, bool) {
 	name := call.Function.Name
 
 	if a.toolRegistry == nil {
-		return tools.FormatError(fmt.Errorf("tool registry is disabled"))
+		return tools.FormatError(fmt.Errorf("tool registry is disabled")), false
 	}
 	tool, ok := a.toolRegistry.Get(name)
 	if !ok {
-		return tools.FormatError(fmt.Errorf("unknown tool: %s", name))
+		return tools.FormatError(fmt.Errorf("unknown tool: %s", name)), false
 	}
 
 	args := strings.TrimSpace(call.Function.Arguments)
 	if args == "" {
 		args = "{}"
 	}
+	raw := json.RawMessage(args)
 
 	toolCtx, cancel := context.WithTimeout(ctx, a.toolTimeout)
 	defer cancel()
-	output, err := tool.Execute(toolCtx, json.RawMessage(args))
+
+	if streaming, ok := tool.(tools.StreamingTool); ok {
+		return a.executeStreamingTool(streaming, toolCtx, raw, name, args, logs)
+	}
+
+	output, err := tool.Execute(toolCtx, raw)
 	if err != nil {
 		if output != "" {
-			return output + "\n" + tools.FormatError(err)
+			return output + "\n" + tools.FormatError(err), false
 		}
-		return tools.FormatError(err)
+		return tools.FormatError(err), false
 	}
-	return output
+	return output, false
+}
+
+func (a *Agent) executeStreamingTool(streaming tools.StreamingTool, ctx context.Context, raw json.RawMessage, name, args string, logs io.Writer) (string, bool) {
+	var streamed int
+	emit := func(chunk string) error {
+		if chunk == "" || logs == nil {
+			return nil
+		}
+		if streamed >= toolStreamLogLimit {
+			return nil
+		}
+		if len(chunk) > toolStreamLogLimit-streamed {
+			chunk = chunk[:toolStreamLogLimit-streamed]
+			streamed = toolStreamLogLimit
+			_, err := fmt.Fprintf(logs, "\n[tool] %s %s\n%s\n%s\n[streaming output truncated after %d bytes]", name, compact(args, toolLogArgumentLimit), toolLogOutputMarker, chunk, toolStreamLogLimit)
+			return err
+		}
+		streamed += len(chunk)
+		_, err := fmt.Fprintf(logs, "\n[tool] %s %s\n%s\n%s", name, compact(args, toolLogArgumentLimit), toolLogOutputMarker, chunk)
+		return err
+	}
+
+	output, err := streaming.ExecuteStream(ctx, raw, emit)
+	if logs != nil {
+		if err != nil {
+			_ = emit("\n" + tools.FormatError(err))
+		} else if streamed == 0 {
+			_ = emit("(no output)")
+		}
+	}
+	if err != nil {
+		if output != "" {
+			return output + "\n" + tools.FormatError(err), true
+		}
+		return tools.FormatError(err), true
+	}
+	return output, true
 }
 
 func (a *Agent) hasMutatingCall(calls []llm.ToolCall) bool {
@@ -498,9 +594,9 @@ func (a *Agent) executeTools(ctx context.Context, calls []llm.ToolCall, logs io.
 	if a.hasMutatingCall(calls) {
 		// Sequential: avoid racing mutating tools.
 		for i, call := range calls {
-			output := a.executeTool(ctx, call)
+			output, streamed := a.executeTool(ctx, call, logs)
 			results[i] = toolExecutionResult{call: call, output: output}
-			if logs != nil {
+			if logs != nil && !streamed {
 				logToolOutput(logs, call, output)
 			}
 		}
@@ -514,9 +610,9 @@ func (a *Agent) executeTools(ctx context.Context, calls []llm.ToolCall, logs io.
 		wg.Add(1)
 		go func(idx int, c llm.ToolCall) {
 			defer wg.Done()
-			output := a.executeTool(ctx, c)
+			output, streamed := a.executeTool(ctx, c, logs)
 			results[idx] = toolExecutionResult{call: c, output: output}
-			if logs != nil {
+			if logs != nil && !streamed {
 				mu.Lock()
 				logToolOutput(logs, c, output)
 				mu.Unlock()

internal/agent/agent_test.go

@@ -87,6 +87,86 @@ func TestAgentRunsToolLoop(t *testing.T) {
 	}
 }
 
+type streamToolProvider struct {
+	calls int
+	t     *testing.T
+}
+
+func (p *streamToolProvider) ChatStream(ctx context.Context, req llm.ChatRequest, emit func(llm.StreamEvent) error) error {
+	p.calls++
+	switch p.calls {
+	case 1:
+		return emit(llm.StreamEvent{ToolCalls: []llm.ToolCallDelta{
+			{Index: 0, ID: "call_1", Type: "function", Name: "test_stream", Arguments: `{}`},
+		}})
+	case 2:
+		last := req.Messages[len(req.Messages)-1]
+		if last.Role != llm.RoleTool || last.Content != "full result" {
+			p.t.Fatalf("last message = %#v", last)
+		}
+		return emit(llm.StreamEvent{Content: "done"})
+	default:
+		p.t.Fatalf("unexpected call count %d", p.calls)
+		return nil
+	}
+}
+
+type streamEchoTool struct{}
+
+func (streamEchoTool) Definition() llm.Tool {
+	return llm.Tool{
+		Type: "function",
+		Function: llm.ToolFunction{
+			Name:        "test_stream",
+			Description: "streaming test",
+			Parameters:  json.RawMessage(`{"type":"object"}`),
+		},
+	}
+}
+
+func (streamEchoTool) Execute(ctx context.Context, raw json.RawMessage) (string, error) {
+	return "full result", nil
+}
+
+func (streamEchoTool) ExecuteStream(ctx context.Context, raw json.RawMessage, emit func(string) error) (string, error) {
+	if err := emit("one\n"); err != nil {
+		return "", err
+	}
+	if err := emit("two\n"); err != nil {
+		return "", err
+	}
+	return "full result", nil
+}
+
+func TestAgentStreamsToolOutputToLogs(t *testing.T) {
+	provider := &streamToolProvider{t: t}
+	a := New(Options{
+		Provider:     provider,
+		Model:        "test",
+		SystemPrompt: "system",
+		ToolRegistry: tools.NewRegistry(streamEchoTool{}),
+	})
+
+	var out strings.Builder
+	var logs strings.Builder
+	if err := a.RunTurn(context.Background(), "stream", &out, &logs); err != nil {
+		t.Fatal(err)
+	}
+	if out.String() != "done" {
+		t.Fatalf("out = %q", out.String())
+	}
+	logText := logs.String()
+	if strings.Count(logText, "[output]") != 2 {
+		t.Fatalf("logs = %q", logText)
+	}
+	if !strings.Contains(logText, "one\n") || !strings.Contains(logText, "two\n") {
+		t.Fatalf("logs missing streamed chunks: %q", logText)
+	}
+	if strings.Contains(logText, "[output]\nfull result") {
+		t.Fatalf("logs contain duplicate full output: %q", logText)
+	}
+}
+
 type contentProvider struct {
 	called bool
 	text   string
@@ -145,6 +225,9 @@ func TestAgentCompact(t *testing.T) {
 	if messages[len(messages)-1].Content != "recent-answer-2" {
 		t.Fatalf("last recent message = %#v", messages[len(messages)-1])
 	}
+	if messages[2].Content != "recent-user-0" {
+		t.Fatalf("first retained recent message = %#v", messages[2])
+	}
 	if usage := a.LastUsage(); usage == nil || usage.TotalTokens != 12 {
 		t.Fatalf("usage = %#v", usage)
 	}
@@ -217,6 +300,48 @@ func longConversation() []llm.Message {
 	return messages
 }
 
+func manyTurnConversation(turns int) []llm.Message {
+	messages := []llm.Message{{Role: llm.RoleSystem, Content: "system"}}
+	for i := 0; i < turns; i++ {
+		messages = append(messages,
+			llm.Message{Role: llm.RoleUser, Content: fmt.Sprintf("user-%02d", i)},
+			llm.Message{Role: llm.RoleAssistant, Content: fmt.Sprintf("assistant-%02d", i)},
+		)
+	}
+	return messages
+}
+
+func TestAgentCompactKeepsTwentyPercentRecentContext(t *testing.T) {
+	provider := &compactProvider{}
+	a := New(Options{Provider: provider, Model: "test", SystemPrompt: "system"})
+	a.SetMessages(manyTurnConversation(25))
+
+	if err := a.Compact(context.Background()); err != nil {
+		t.Fatal(err)
+	}
+
+	messages := a.Messages()
+	// 1 system + 1 summary + 10 retained recent messages (5 user/assistant turns)
+	if len(messages) != 12 {
+		t.Fatalf("messages = %d, want 12; contents: %v", len(messages), joinMessageContents(messages))
+	}
+	if provider.summaryCalls != 1 {
+		t.Fatalf("summaryCalls = %d, want 1", provider.summaryCalls)
+	}
+	if messages[2].Content != "user-20" {
+		t.Fatalf("first retained message = %#v, want user-20", messages[2])
+	}
+	if messages[len(messages)-1].Content != "assistant-24" {
+		t.Fatalf("last retained message = %#v, want assistant-24", messages[len(messages)-1])
+	}
+	if !strings.Contains(provider.lastSummary, "user-19") {
+		t.Fatalf("summary should contain user-19 (part of summarized 80%%): %q", provider.lastSummary)
+	}
+	if strings.Contains(provider.lastSummary, "user-20") {
+		t.Fatalf("summary should not contain user-20 (part of retained 20%%): %q", provider.lastSummary)
+	}
+}
+
 func joinMessageContents(messages []llm.Message) string {
 	var b strings.Builder
 	for _, msg := range messages {

internal/app/app.go

@@ -0,0 +1,171 @@
+package app
+
+import (
+	"context"
+	"errors"
+	"flag"
+	"fmt"
+	"net/http"
+	"os"
+	"os/signal"
+	"path/filepath"
+	"strings"
+
+	"agentu/internal/agent"
+	"agentu/internal/session"
+	"agentu/internal/tools"
+	"agentu/internal/tui"
+	"agentu/pkg/config"
+	"agentu/pkg/llm"
+)
+
+type cliOptions struct {
+	configPath string
+	yolo       bool
+	plain      bool
+	themeName  string
+	resumeID   string
+}
+
+func Run() error {
+	opts := parseCLIOptions()
+
+	themeMode, err := tui.ParseThemeMode(opts.themeName)
+	if err != nil {
+		return err
+	}
+
+	cfg, err := loadConfig(opts.configPath)
+	if err != nil {
+		return err
+	}
+
+	providerName, providerConfig, provider := defaultProvider(cfg, http.DefaultClient)
+	assistant := newAssistant(cfg, providerName, providerConfig, provider, opts.yolo, http.DefaultClient)
+	modelManager, err := newSessionManager(cfg, assistant, http.DefaultClient)
+	if err != nil {
+		return err
+	}
+
+	if err := initializeSession(modelManager, opts.resumeID); err != nil {
+		return err
+	}
+
+	ctx, stop := signal.NotifyContext(context.Background(), os.Interrupt)
+	defer stop()
+
+	return runInterface(ctx, assistant, modelManager, providerConfig, cfg.Agent.WorkingDir, opts, themeMode)
+}
+
+func parseCLIOptions() cliOptions {
+	configPath := flag.String("config", config.DefaultPath, "path to YAML config")
+	yolo := flag.Bool("yolo", false, "also enable automatic file writes and shell execution")
+	plain := flag.Bool("plain", false, "use the simple line-based REPL instead of the TUI")
+	themeName := flag.String("theme", "light", "TUI theme: light or dark")
+	resumeID := flag.String("resume", "", "resume a specific session by ID")
+	flag.Parse()
+
+	return cliOptions{
+		configPath: *configPath,
+		yolo:       *yolo,
+		plain:      *plain,
+		themeName:  *themeName,
+		resumeID:   *resumeID,
+	}
+}
+
+func loadConfig(configPath string) (*config.Config, error) {
+	cfg, err := config.Load(configPath)
+	if err == nil {
+		return cfg, nil
+	}
+	if !errors.Is(err, os.ErrNotExist) {
+		return nil, err
+	}
+	if bootstrapErr := bootstrapConfig(configPath); bootstrapErr != nil {
+		return nil, bootstrapErr
+	}
+	return nil, fmt.Errorf("created default config at %s; edit it to set your API key and provider, then restart agentu", configPath)
+}
+
+func defaultProvider(cfg *config.Config, httpClient *http.Client) (string, config.ProviderConfig, llm.Provider) {
+	providerName := cfg.DefaultProviderName()
+	providerConfig := cfg.Providers[providerName]
+	provider := newProviderClient(providerConfig, httpClient)
+	return providerName, providerConfig, provider
+}
+
+func newProviderClient(providerConfig config.ProviderConfig, httpClient *http.Client) llm.Provider {
+	if providerConfig.APIType == config.APITypeResponses {
+		return llm.NewOpenAIResponsesClient(providerConfig.BaseURL, providerConfig.APIKey, httpClient)
+	}
+	return llm.NewOpenAICompatibleClient(providerConfig.BaseURL, providerConfig.APIKey, httpClient)
+}
+
+func newAssistant(cfg *config.Config, providerName string, providerConfig config.ProviderConfig, provider llm.Provider, yolo bool, httpClient *http.Client) *agent.Agent {
+	return agent.New(agent.Options{
+		Provider:         provider,
+		ProviderName:     providerName,
+		Model:            providerConfig.Model,
+		Thinking:         providerConfig.Thinking,
+		ThinkingParam:    providerConfig.ThinkingParam,
+		ThinkingEnabled:  providerConfig.ThinkingConfigured,
+		SystemPrompt:     systemPrompt(cfg, yolo),
+		ToolRegistry:     newToolRegistry(cfg.Agent.WorkingDir, yolo, httpClient),
+		ToolTimeout:      cfg.Agent.ToolTimeout,
+		MaxContextTokens: maxContextTokens(cfg, providerConfig),
+	})
+}
+
+func systemPrompt(cfg *config.Config, yolo bool) string {
+	instructions := tools.AgentInstructions(cfg.Agent.WorkingDir, yolo)
+	return strings.TrimSpace(cfg.Agent.SystemPrompt + "\n\n" + instructions)
+}
+
+func maxContextTokens(cfg *config.Config, providerConfig config.ProviderConfig) int {
+	if cfg.Agent.MaxContextTokens != 0 {
+		return cfg.Agent.MaxContextTokens
+	}
+	return providerConfig.ContextTokens
+}
+
+func newToolRegistry(workingDir string, yolo bool, httpClient *http.Client) *tools.Registry {
+	var registry *tools.Registry
+	if yolo {
+		registry = tools.Builtins(workingDir)
+	} else {
+		registry = tools.ReadOnlyBuiltins(workingDir)
+	}
+	registry.Register(tools.NewBuiltinSearchTool(httpClient))
+	registry.Register(tools.NewBuiltinFetchTool(httpClient))
+	return registry
+}
+
+func newSessionManager(cfg *config.Config, assistant *agent.Agent, httpClient *http.Client) (*session.Manager, error) {
+	home, err := os.UserHomeDir()
+	if err != nil {
+		return nil, fmt.Errorf("resolve home dir: %w", err)
+	}
+	sessionsDir := filepath.Join(home, ".agentu", "sessions")
+	store, err := session.NewStore(sessionsDir)
+	if err != nil {
+		return nil, fmt.Errorf("init session store: %w", err)
+	}
+	return session.NewManager(cfg, assistant, httpClient, store)
+}
+
+func runInterface(ctx context.Context, assistant *agent.Agent, modelManager *session.Manager, providerConfig config.ProviderConfig, workingDir string, opts cliOptions, themeMode tui.ThemeMode) error {
+	if opts.plain {
+		return repl(ctx, assistant, modelManager)
+	}
+
+	err := tui.Run(ctx, assistant, tui.Options{
+		ModelName:    providerConfig.Model,
+		Yolo:         opts.yolo,
+		ThemeMode:    themeMode,
+		ModelManager: modelManager,
+		WorkingDir:   workingDir,
+	})
+	printExitSession(modelManager)
+	return err
+}

internal/app/bootstrap.go

@@ -0,0 +1,38 @@
+package app
+
+import (
+	"fmt"
+	"os"
+	"path/filepath"
+
+	"agentu/pkg/config"
+)
+
+const defaultConfigContent = `# agentu configuration
+# Replace the placeholder values below with your actual provider settings.
+# See agentu.example.yaml for the full reference.
+providers:
+  default:
+    base_url: https://api.openai.com
+    api_key: YOUR_API_KEY_HERE
+    models:
+      - id: gpt-4o
+        context: 128000
+      - id: gpt-4o-mini
+        context: 128000
+`
+
+func bootstrapConfig(configPath string) error {
+	resolved, err := config.ResolvePath(configPath)
+	if err != nil {
+		return fmt.Errorf("resolve config path: %w", err)
+	}
+	dir := filepath.Dir(resolved)
+	if err := os.MkdirAll(dir, 0o755); err != nil {
+		return fmt.Errorf("create config directory: %w", err)
+	}
+	if err := os.WriteFile(resolved, []byte(defaultConfigContent), 0o644); err != nil {
+		return fmt.Errorf("write default config: %w", err)
+	}
+	return nil
+}

internal/app/main_test.go

@@ -1,4 +1,4 @@
-package main
+package app
 
 import (
 	"context"

internal/app/repl.go

@@ -0,0 +1,60 @@
+package app
+
+import (
+	"bufio"
+	"context"
+	"fmt"
+	"os"
+	"strings"
+
+	"agentu/internal/agent"
+	"agentu/internal/session"
+)
+
+func repl(ctx context.Context, assistant *agent.Agent, modelManager *session.Manager) error {
+	scanner := bufio.NewScanner(os.Stdin)
+	scanner.Buffer(make([]byte, 0, 64*1024), 1024*1024)
+
+	fmt.Println("agentu")
+	fmt.Println("Type /exit to quit, /clear to reset context, /compact to compress context.")
+	for {
+		fmt.Print("> ")
+		if !scanner.Scan() {
+			if err := scanner.Err(); err != nil {
+				return err
+			}
+			fmt.Println()
+			printExitSession(modelManager)
+			return nil
+		}
+
+		input := strings.TrimSpace(scanner.Text())
+		switch input {
+		case "":
+			continue
+		case "/exit", "/quit":
+			printExitSession(modelManager)
+			return nil
+		case "/clear":
+			assistant.Clear()
+			fmt.Println("context cleared")
+			continue
+		case "/compact":
+			if err := assistant.Compact(ctx); err != nil {
+				fmt.Fprintln(os.Stderr, "compact error:", err)
+			} else {
+				fmt.Println("context compacted")
+				if modelManager != nil {
+					_ = modelManager.Save()
+				}
+			}
+			continue
+		}
+
+		if err := assistant.RunTurn(ctx, input, os.Stdout, os.Stderr); err != nil {
+			fmt.Fprintln(os.Stderr, "\nerror:", err)
+		}
+		_ = modelManager.Save()
+		fmt.Println()
+	}
+}

internal/app/session_startup.go

@@ -0,0 +1,38 @@
+package app
+
+import (
+	"fmt"
+	"os"
+
+	"agentu/internal/session"
+)
+
+func initializeSession(modelManager *session.Manager, resumeID string) error {
+	if resumeID != "" {
+		if err := modelManager.Resume(resumeID); err != nil {
+			return fmt.Errorf("resume session: %w", err)
+		}
+		return nil
+	}
+	if _, err := modelManager.NewSession(); err != nil {
+		return fmt.Errorf("start new session: %w", err)
+	}
+	return nil
+}
+
+func printExitSession(m *session.Manager) {
+	if m == nil {
+		return
+	}
+	_ = m.Save()
+	id := m.CurrentSessionID()
+	name := m.CurrentSessionName()
+	if id != "" {
+		if name != "" && name != id {
+			fmt.Fprintf(os.Stderr, "session: %s (%s)\n", id, name)
+		} else {
+			fmt.Fprintf(os.Stderr, "session: %s\n", id)
+		}
+		fmt.Fprintf(os.Stderr, "resume:  agentu --resume %s\n", id)
+	}
+}

internal/session/manager.go

@@ -68,11 +68,29 @@ func (m *Manager) Resume(id string) error {
 		return err
 	}
 	m.currentSession = sess
+	m.restoreRuntime(sess)
 	m.agent.SetMessages(sess.Messages)
 	m.agent.SetLastUsage(sess.LastUsage)
 	return nil
 }
 
+// restoreRuntime switches the manager and agent back to the provider, model,
+// and thinking level recorded on a resumed session. Values that are no longer
+// valid for the current config are skipped and the defaults remain active.
+func (m *Manager) restoreRuntime(sess *Session) {
+	if sess.Provider != "" && sess.Provider != m.providerName {
+		if _, ok := m.cfg.Providers[sess.Provider]; ok {
+			_, _ = m.SetProvider(sess.Provider)
+		}
+	}
+	if sess.Model != "" && sess.Model != m.provider.Model {
+		_, _ = m.SetModel(sess.Model)
+	}
+	if sess.Thinking != "" && sess.Thinking != m.provider.Thinking {
+		_, _ = m.SetThinking(sess.Thinking)
+	}
+}
+
 // Save persists the current session to disk.
 func (m *Manager) Save() error {
 	if m.store == nil || m.currentSession == nil {
@@ -82,6 +100,8 @@ func (m *Manager) Save() error {
 	m.currentSession.LastUsage = m.agent.LastUsage()
 	m.currentSession.Provider = m.providerName
 	m.currentSession.Model = m.provider.Model
+	m.currentSession.APIType = m.provider.APIType
+	m.currentSession.Thinking = m.provider.Thinking
 	// Auto-name if empty
 	if m.currentSession.Name == "" {
 		m.currentSession.Name = DefaultName(m.currentSession.Messages)
@@ -150,6 +170,8 @@ func (m *Manager) NewSession() (string, error) {
 			m.currentSession.Messages = msgs
 			m.currentSession.Provider = m.providerName
 			m.currentSession.Model = m.provider.Model
+			m.currentSession.APIType = m.provider.APIType
+			m.currentSession.Thinking = m.provider.Thinking
 			if m.currentSession.Name == "" {
 				m.currentSession.Name = DefaultName(msgs)
 			}
@@ -204,10 +226,18 @@ func (m *Manager) CurrentThinking() string {
 	return m.provider.Thinking
 }
 
+func (m *Manager) CurrentAPIType() string {
+	if m.provider.APIType == "" {
+		return config.APITypeChat
+	}
+	return m.provider.APIType
+}
+
 func (m *Manager) Info() string {
 	var b strings.Builder
 	fmt.Fprintf(&b, "provider: %s\n", m.CurrentProvider())
 	fmt.Fprintf(&b, "model: %s\n", m.CurrentModel())
+	fmt.Fprintf(&b, "api: %s\n", m.CurrentAPIType())
 	fmt.Fprintf(&b, "thinking: %s\n", m.CurrentThinking())
 	fmt.Fprintf(&b, "providers: %s\n", strings.Join(m.providerNames(), ", "))
 	fmt.Fprintf(&b, "models: %s\n", strings.Join(m.provider.Models, ", "))
@@ -223,7 +253,7 @@ func (m *Manager) SetProvider(name string) (string, error) {
 	}
 	m.providerName = name
 	m.provider = provider
-	m.syncAgent(llm.NewOpenAICompatibleClient(provider.BaseURL, provider.APIKey, m.httpClient))
+	m.syncAgent(providerClient(provider, m.httpClient))
 	return "Switched provider.\n" + m.summary(), nil
 }
 
@@ -249,7 +279,7 @@ func (m *Manager) SetThinking(level string) (string, error) {
 	m.provider.Thinking = level
 	m.provider.ThinkingConfigured = true
 	if m.provider.ThinkingParam == "" {
-		m.provider.ThinkingParam = "thinking"
+		m.provider.ThinkingParam = config.DefaultThinkingParam(m.provider.APIType)
 	}
 	m.cfg.Providers[m.providerName] = m.provider
 	m.syncAgent(nil)
@@ -257,7 +287,14 @@ func (m *Manager) SetThinking(level string) (string, error) {
 }
 
 func (m *Manager) summary() string {
-	return fmt.Sprintf("provider: %s\nmodel: %s\nthinking: %s", m.CurrentProvider(), m.CurrentModel(), m.CurrentThinking())
+	return fmt.Sprintf("provider: %s\nmodel: %s\napi: %s\nthinking: %s", m.CurrentProvider(), m.CurrentModel(), m.CurrentAPIType(), m.CurrentThinking())
+}
+
+func providerClient(provider config.ProviderConfig, httpClient *http.Client) llm.Provider {
+	if provider.APIType == config.APITypeResponses {
+		return llm.NewOpenAIResponsesClient(provider.BaseURL, provider.APIKey, httpClient)
+	}
+	return llm.NewOpenAICompatibleClient(provider.BaseURL, provider.APIKey, httpClient)
 }
 
 func (m *Manager) maxContextTokens() int {
@@ -288,9 +325,13 @@ func (m *Manager) providerNames() []string {
 	return names
 }
 
+func (m *Manager) AvailableModels() []string {
+	return append([]string(nil), m.provider.Models...)
+}
+
 func contains(values []string, value string) bool {
-	for _, item := range values {
-		if item == value {
+	for _, v := range values {
+		if v == value {
 			return true
 		}
 	}

internal/session/manager_test.go

@@ -2,6 +2,8 @@ package session
 
 import (
 	"context"
+	"net/http"
+	"net/http/httptest"
 	"path/filepath"
 	"strings"
 	"testing"
@@ -109,6 +111,76 @@ func TestManagerRejectsUnknownModelAndThinking(t *testing.T) {
 	}
 }
 
+func TestManagerSwitchesToResponsesProvider(t *testing.T) {
+	var hitPath string
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		hitPath = r.URL.Path
+		w.Header().Set("Content-Type", "text/event-stream")
+		_, _ = w.Write([]byte(`data: {"type":"response.output_text.delta","item_id":"msg_1","output_index":0,"content_index":0,"delta":"ok"}` + "\n\n"))
+		_, _ = w.Write([]byte(`data: {"type":"response.completed","response":{"id":"resp_1","status":"completed"}}` + "\n\n"))
+	}))
+	defer server.Close()
+
+	cfg := &config.Config{
+		Providers: map[string]config.ProviderConfig{
+			"chat": {
+				Name:    "chat",
+				APIType: config.APITypeChat,
+				BaseURL: "https://chat.example.com",
+				APIKey:  "sk-test",
+				Model:   "model-a",
+				Models:  []string{"model-a"},
+			},
+			"responses": {
+				Name:    "responses",
+				APIType: config.APITypeResponses,
+				BaseURL: server.URL,
+				APIKey:  "sk-test",
+				Model:   "model-b",
+				Models:  []string{"model-b"},
+			},
+		},
+	}
+	assistant := agent.New(agent.Options{Provider: &recordingProvider{}, Model: "model-a"})
+	store, err := NewStore(filepath.Join(t.TempDir(), "sessions"))
+	if err != nil {
+		t.Fatal(err)
+	}
+	manager, err := NewManager(cfg, assistant, server.Client(), store)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	if _, err := manager.SetProvider("responses"); err != nil {
+		t.Fatal(err)
+	}
+	if manager.CurrentAPIType() != config.APITypeResponses {
+		t.Fatalf("api type = %q", manager.CurrentAPIType())
+	}
+	var out strings.Builder
+	if err := assistant.RunTurn(context.Background(), "hi", &out, nil); err != nil {
+		t.Fatal(err)
+	}
+	if hitPath != "/v1/responses" {
+		t.Fatalf("path = %q, want /v1/responses", hitPath)
+	}
+	if out.String() != "ok" {
+		t.Fatalf("output = %q", out.String())
+	}
+}
+
+func TestManagerExposesModelCommandCandidates(t *testing.T) {
+	manager, _, _ := testManager(t)
+	models := manager.AvailableModels()
+	if len(models) != 2 || models[0] != "model-a" || models[1] != "model-b" {
+		t.Fatalf("AvailableModels() = %v", models)
+	}
+	models[0] = "mutated"
+	if got := manager.AvailableModels()[0]; got != "model-a" {
+		t.Fatalf("mutation leaked: %q", got)
+	}
+}
+
 func TestManagerSaveAndResume(t *testing.T) {
 	manager, assistant, _ := testManager(t)
 
@@ -331,3 +403,132 @@ func TestManagerUsesModelContextForAgentCompaction(t *testing.T) {
 		t.Fatalf("answerCalls = %d, want 1", provider.answerCalls)
 	}
 }
+
+func TestManagerResumeRestoresRuntime(t *testing.T) {
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		w.Header().Set("Content-Type", "text/event-stream")
+		_, _ = w.Write([]byte(`data: {"type":"response.output_text.delta","item_id":"msg_1","output_index":0,"content_index":0,"delta":"ok"}` + "\n\n"))
+		_, _ = w.Write([]byte(`data: {"type":"response.completed","response":{"id":"resp_1","status":"completed"}}` + "\n\n"))
+	}))
+	defer server.Close()
+
+	cfg := &config.Config{
+		Providers: map[string]config.ProviderConfig{
+			"chat": {
+				Name:               "chat",
+				APIType:            config.APITypeChat,
+				BaseURL:            "https://chat.example.com",
+				APIKey:             "sk-test",
+				Model:              "model-a",
+				Models:             []string{"model-a"},
+				Thinking:           "none",
+				ThinkingParam:      "thinking",
+				ThinkingConfigured: true,
+			},
+			"resp": {
+				Name:               "resp",
+				APIType:            config.APITypeResponses,
+				BaseURL:            server.URL,
+				APIKey:             "sk-test",
+				Model:              "m1",
+				Models:             []string{"m1", "m2"},
+				Thinking:           "none",
+				ThinkingParam:      "reasoning",
+				ThinkingConfigured: true,
+			},
+		},
+	}
+	assistant := agent.New(agent.Options{
+		Provider:        &recordingProvider{},
+		Model:           "model-a",
+		Thinking:        "none",
+		ThinkingParam:   "thinking",
+		ThinkingEnabled: true,
+	})
+	store, err := NewStore(filepath.Join(t.TempDir(), "sessions"))
+	if err != nil {
+		t.Fatal(err)
+	}
+	manager, err := NewManager(cfg, assistant, server.Client(), store)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	// Build a session on the responses provider with a custom model and thinking.
+	manager.newSession()
+	if _, err := manager.SetProvider("resp"); err != nil {
+		t.Fatal(err)
+	}
+	if _, err := manager.SetModel("m2"); err != nil {
+		t.Fatal(err)
+	}
+	if _, err := manager.SetThinking("xhigh"); err != nil {
+		t.Fatal(err)
+	}
+	if err := manager.Save(); err != nil {
+		t.Fatal(err)
+	}
+	id := manager.CurrentSessionID()
+
+	// Drift away from the saved runtime.
+	if _, err := manager.SetProvider("chat"); err != nil {
+		t.Fatal(err)
+	}
+	if _, err := manager.SetModel("model-a"); err != nil {
+		t.Fatal(err)
+	}
+	if _, err := manager.SetThinking("none"); err != nil {
+		t.Fatal(err)
+	}
+
+	if err := manager.Resume(id); err != nil {
+		t.Fatal(err)
+	}
+	if got := manager.CurrentProvider(); got != "resp" {
+		t.Fatalf("provider = %q, want resp", got)
+	}
+	if got := manager.CurrentModel(); got != "m2" {
+		t.Fatalf("model = %q, want m2", got)
+	}
+	if got := manager.CurrentThinking(); got != "xhigh" {
+		t.Fatalf("thinking = %q, want xhigh", got)
+	}
+	if got := manager.CurrentAPIType(); got != config.APITypeResponses {
+		t.Fatalf("api type = %q, want responses", got)
+	}
+	runtime := assistant.RuntimeInfo()
+	if runtime.ProviderName != "resp" || runtime.Model != "m2" || runtime.Thinking != "xhigh" {
+		t.Fatalf("agent runtime = %#v", runtime)
+	}
+}
+
+func TestManagerResumeFallsBackWhenRuntimeMissing(t *testing.T) {
+	manager, assistant, _ := testManager(t)
+
+	sess := &Session{
+		ID:       GenerateID(),
+		Provider: "ghost",
+		Model:    "ghost-model",
+		Thinking: "xhigh",
+		Messages: []llm.Message{{Role: llm.RoleUser, Content: "hi"}},
+	}
+	if err := manager.store.Save(sess); err != nil {
+		t.Fatal(err)
+	}
+	if err := manager.Resume(sess.ID); err != nil {
+		t.Fatal(err)
+	}
+	if got := manager.CurrentProvider(); got != "one" {
+		t.Fatalf("provider = %q, want default one", got)
+	}
+	if got := manager.CurrentModel(); got != "model-a" {
+		t.Fatalf("model = %q, want default model-a", got)
+	}
+	if got := manager.CurrentThinking(); got != "xhigh" {
+		t.Fatalf("thinking = %q, want restored xhigh", got)
+	}
+	runtime := assistant.RuntimeInfo()
+	if runtime.ProviderName != "one" || runtime.Model != "model-a" {
+		t.Fatalf("agent runtime = %#v", runtime)
+	}
+}

internal/session/store.go

@@ -22,6 +22,8 @@ type Session struct {
 	UpdatedAt time.Time     `json:"updated_at"`
 	Provider  string        `json:"provider"`
 	Model     string        `json:"model"`
+	APIType   string        `json:"api_type,omitempty"`
+	Thinking  string        `json:"thinking,omitempty"`
 	Messages  []llm.Message `json:"messages"`
 	LastUsage *llm.Usage    `json:"last_usage,omitempty"`
 }

internal/tools/shell.go

@@ -1,6 +1,7 @@
 package tools
 
 import (
+	"bytes"
 	"context"
 	"encoding/json"
 	"errors"
@@ -8,6 +9,7 @@ import (
 	"os"
 	"os/exec"
 	"strings"
+	"sync"
 
 	"agentu/pkg/llm"
 )
@@ -46,6 +48,10 @@ type shellRunArgs struct {
 }
 
 func (t *ShellRunTool) Execute(ctx context.Context, raw json.RawMessage) (string, error) {
+	return t.ExecuteStream(ctx, raw, nil)
+}
+
+func (t *ShellRunTool) ExecuteStream(ctx context.Context, raw json.RawMessage, emit func(string) error) (string, error) {
 	var args shellRunArgs
 	if err := json.Unmarshal(raw, &args); err != nil {
 		return "", fmt.Errorf("parse arguments: %w", err)
@@ -66,16 +72,45 @@ func (t *ShellRunTool) Execute(ctx context.Context, raw json.RawMessage) (string
 	cmd := exec.CommandContext(ctx, shell, "-lc", args.Command)
 	cmd.Dir = workdir
 
-	output, err := cmd.CombinedOutput()
-	text := FormatResult(string(output))
+	var mu sync.Mutex
+	var combined bytes.Buffer
+	writer := &shellStreamWriter{mu: &mu, buf: &combined, emit: emit}
+	cmd.Stdout = writer
+	cmd.Stderr = writer
+
+	runErr := cmd.Run()
+	text := FormatResult(combined.String())
 	if errors.Is(ctx.Err(), context.DeadlineExceeded) {
 		return text, fmt.Errorf("command timed out: %w", ctx.Err())
 	}
-	if err != nil {
+	if runErr != nil {
 		if text != "" {
-			return text, fmt.Errorf("command failed: %w", err)
+			return text, fmt.Errorf("command failed: %w", runErr)
 		}
-		return "", fmt.Errorf("command failed: %w", err)
+		return "", fmt.Errorf("command failed: %w", runErr)
 	}
 	return text, nil
 }
+
+// shellStreamWriter captures command output for the final result while
+// forwarding every write to the streaming callback as it happens.
+type shellStreamWriter struct {
+	mu   *sync.Mutex
+	buf  *bytes.Buffer
+	emit func(string) error
+}
+
+func (w *shellStreamWriter) Write(p []byte) (int, error) {
+	w.mu.Lock()
+	n, writeErr := w.buf.Write(p)
+	w.mu.Unlock()
+	if writeErr != nil {
+		return n, writeErr
+	}
+	if w.emit != nil {
+		if err := w.emit(string(p)); err != nil {
+			return 0, err
+		}
+	}
+	return len(p), nil
+}

internal/tools/shell_test.go

@@ -0,0 +1,71 @@
+package tools
+
+import (
+	"context"
+	"encoding/json"
+	"strings"
+	"sync"
+	"testing"
+)
+
+func TestShellRunStreamsOutputWhileRunning(t *testing.T) {
+	tool := NewShellRunTool(t.TempDir())
+	var mu sync.Mutex
+	var chunks []string
+	output, err := tool.ExecuteStream(context.Background(), json.RawMessage(`{"command":"printf 'a\\n'; sleep 0.05; printf 'b\\n'"}`), func(chunk string) error {
+		mu.Lock()
+		chunks = append(chunks, chunk)
+		mu.Unlock()
+		return nil
+	})
+	if err != nil {
+		t.Fatal(err)
+	}
+	if !strings.Contains(output, "a") || !strings.Contains(output, "b") {
+		t.Fatalf("output = %q", output)
+	}
+	mu.Lock()
+	joined := strings.Join(chunks, "")
+	mu.Unlock()
+	if len(chunks) == 0 {
+		t.Fatal("expected at least one streamed chunk")
+	}
+	if !strings.Contains(joined, "a") || !strings.Contains(joined, "b") {
+		t.Fatalf("streamed chunks = %q", joined)
+	}
+}
+
+func TestShellRunExecuteCollectsFullOutput(t *testing.T) {
+	tool := NewShellRunTool(t.TempDir())
+	output, err := tool.Execute(context.Background(), json.RawMessage(`{"command":"printf 'hello world'"}`))
+	if err != nil {
+		t.Fatal(err)
+	}
+	if strings.TrimSpace(output) != "hello world" {
+		t.Fatalf("output = %q", output)
+	}
+}
+
+func TestShellRunStreamsStderrAndStdout(t *testing.T) {
+	tool := NewShellRunTool(t.TempDir())
+	var mu sync.Mutex
+	var chunks []string
+	output, err := tool.ExecuteStream(context.Background(), json.RawMessage(`{"command":"echo out; echo err >&2"}`), func(chunk string) error {
+		mu.Lock()
+		chunks = append(chunks, chunk)
+		mu.Unlock()
+		return nil
+	})
+	if err != nil {
+		t.Fatal(err)
+	}
+	if !strings.Contains(output, "out") || !strings.Contains(output, "err") {
+		t.Fatalf("output = %q", output)
+	}
+	mu.Lock()
+	joined := strings.Join(chunks, "")
+	mu.Unlock()
+	if !strings.Contains(joined, "out") || !strings.Contains(joined, "err") {
+		t.Fatalf("streamed chunks = %q", joined)
+	}
+}

internal/tools/tools.go

@@ -19,6 +19,15 @@ type Tool interface {
 	Execute(ctx context.Context, args json.RawMessage) (string, error)
 }
 
+// StreamingTool is optionally implemented by tools that can report partial
+// output while they execute. emit receives chunks of the tool's output as
+// they are produced; the returned string is still the complete result that
+// gets sent back to the model.
+type StreamingTool interface {
+	Tool
+	ExecuteStream(ctx context.Context, args json.RawMessage, emit func(string) error) (string, error)
+}
+
 // MutatingTool is optionally implemented by tools that modify external state.
 // It is used by the agent to determine whether tool calls can run in parallel.
 type MutatingTool interface {

internal/tui/app.go

@@ -50,6 +50,7 @@ type ModelManager interface {
 	Rename(name string) (string, error)
 	Sessions() (string, error)
 	NewSession() (string, error)
+	AvailableModels() []string
 }
 
 func Run(ctx context.Context, assistant *agent.Agent, opts Options) error {
@@ -90,12 +91,13 @@ type model struct {
 	width  int
 	height int
 
-	messages     []message
-	status       string
-	contextUsage string
-	inputHistory []string
-	historyIndex int
-	draftInput   string
+	messages      []message
+	status        string
+	contextUsage  string
+	inputHistory  []string
+	historyIndex  int
+	draftInput    string
+	commandCursor int
 
 	running bool
 	cancel  context.CancelFunc
@@ -133,6 +135,13 @@ type commandMatch struct {
 	score   int
 }
 
+type commandSuggestion struct {
+	label       string
+	description string
+	value       string
+	executable  bool
+}
+
 var slashCommands = []slashCommand{
 	{Name: "/clear", Description: "reset context"},
 	{Name: "/compact", Description: "compress context"},
@@ -145,7 +154,7 @@ var slashCommands = []slashCommand{
 	{Name: "/new", Description: "new session"},
 	{Name: "/status", Description: "show changed files"},
 	{Name: "/diff", Description: "show unstaged diff"},
-	{Name: "/test", Description: "run project tests"},
+	{Name: "/theme", Description: "switch theme light|dark"},
 }
 
 func newModel(ctx context.Context, assistant *agent.Agent, opts Options) model {
@@ -157,7 +166,7 @@ func newModel(ctx context.Context, assistant *agent.Agent, opts Options) model {
 	st := th.styles()
 
 	input := textarea.New()
-	input.Placeholder = "Message agentu..."
+	input.Placeholder = ""
 	input.Prompt = ""
 	input.ShowLineNumbers = false
 	input.MaxHeight = maxInputLines
@@ -197,8 +206,6 @@ func (m model) Init() tea.Cmd {
 	m.syncMessages()
 	return textarea.Blink
 }
-
-// syncMessages rebuilds the TUI display messages from the agent's conversation history.
 func (m *model) syncMessages() {
 	if m.agent == nil {
 		return
@@ -246,6 +253,9 @@ func (m model) Update(msg tea.Msg) (tea.Model, tea.Cmd) {
 		if m.picking {
 			return m.updatePicker(msg)
 		}
+		if handled, next, cmd := m.updateCommandSuggestions(msg); handled {
+			return next, cmd
+		}
 		switch msg.String() {
 		case "ctrl+c":
 			if m.cancel != nil {
@@ -263,6 +273,29 @@ func (m model) Update(msg tea.Msg) (tea.Model, tea.Cmd) {
 				return m, nil
 			}
 			return m.submit()
+		case "up":
+			if m.running {
+				return m, nil
+			}
+			if m.input.Line() == 0 {
+				m.previousInputHistory()
+				m.afterInputChanged()
+				return m, nil
+			}
+		case "down":
+			if m.running {
+				return m, nil
+			}
+			value := m.input.Value()
+			lineCount := 1
+			if value != "" {
+				lineCount = len(strings.Split(value, "\n"))
+			}
+			if m.input.Line() >= lineCount-1 {
+				m.nextInputHistory()
+				m.afterInputChanged()
+				return m, nil
+			}
 		case "ctrl+p":
 			if m.running {
 				return m, nil
@@ -277,7 +310,7 @@ func (m model) Update(msg tea.Msg) (tea.Model, tea.Cmd) {
 			m.nextInputHistory()
 			m.afterInputChanged()
 			return m, nil
-		case "alt+enter", "ctrl+j":
+		case "ctrl+j":
 			return m.insertInputNewline()
 		}
 
@@ -338,7 +371,6 @@ func (m model) Update(msg tea.Msg) (tea.Model, tea.Cmd) {
 	if !m.running {
 		nextInput, cmd := m.input.Update(msg)
 		m.input = nextInput
-		m.afterInputChanged()
 		cmds = append(cmds, cmd)
 	}
 
@@ -356,7 +388,7 @@ func (m model) View() string {
 
 	if m.picking {
 		parts := []string{m.viewport.View()}
-		parts = append(parts, m.pickerView(), m.modelMetaView(), m.footerView())
+		parts = append(parts, m.pickerView(), m.footerView())
 		body := lipgloss.JoinVertical(lipgloss.Left, parts...)
 		return m.styles.App.Width(m.width).Height(m.height).Render(body)
 	}
@@ -365,7 +397,7 @@ func (m model) View() string {
 	if activity := m.activityView(); activity != "" {
 		parts = append(parts, activity)
 	}
-	parts = append(parts, m.inputView(), m.modelMetaView(), m.footerView())
+	parts = append(parts, m.inputView(), m.footerView())
 	body := lipgloss.JoinVertical(lipgloss.Left, parts...)
 	return m.styles.App.Width(m.width).Height(m.height).Render(body)
 }
@@ -389,7 +421,11 @@ func (m *model) submit() (tea.Model, tea.Cmd) {
 		m.status = "ready"
 		m.refreshViewport(true)
 		return *m, nil
-	case "/model", "/compact":
+	case "/model", "/compact", "/theme":
+		return m.runLocalCommand(input)
+	}
+
+	if strings.HasPrefix(input, "!") {
 		return m.runLocalCommand(input)
 	}
 
@@ -438,6 +474,16 @@ func (m *model) insertInputNewline() (tea.Model, tea.Cmd) {
 	m.historyIndex = -1
 	m.draftInput = ""
 	m.afterInputChanged()
+	// The textarea's repositionView was called inside Update with the old
+	// viewport height, so the YOffset may be stale after syncInputHeight
+	// changed the height. Send up then down to force repositionView to
+	// recalculate with the new height.
+	up := tea.KeyMsg{Type: tea.KeyUp}
+	vp, _ := m.input.Update(up)
+	m.input = vp
+	down := tea.KeyMsg{Type: tea.KeyDown}
+	vp, _ = m.input.Update(down)
+	m.input = vp
 	return *m, cmd
 }
 
@@ -467,7 +513,7 @@ func (m *model) appendAssistantChunk(chunk string) {
 
 func (m *model) resize() {
 	width := max(40, m.width)
-	inputWidth := max(20, width-m.styles.InputBox.GetHorizontalFrameSize())
+	inputWidth := max(20, width-m.styles.InputBox.GetHorizontalFrameSize()-m.inputModeWidth()-1)
 	m.input.SetWidth(inputWidth)
 	m.syncInputHeight()
 
@@ -637,15 +683,27 @@ func (m *model) mergeToolLog(text string) bool {
 	if incoming.output == "" {
 		return false
 	}
+	incomingHeader, _, ok := strings.Cut(text, "\n[output]\n")
+	if !ok {
+		return false
+	}
 	for i := len(m.messages) - 1; i >= 0; i-- {
 		if m.messages[i].role != roleTool {
 			continue
 		}
 		existing := parseToolLog(m.messages[i].content)
-		if existing.name == incoming.name && existing.args == incoming.args && existing.output == "" {
-			m.messages[i].content = text
-			return true
+		if existing.name != incoming.name || existing.args != incoming.args {
+			continue
 		}
+		if !strings.Contains(m.messages[i].content, "\n[output]\n") {
+			// Placeholder card from the tool-start log; first chunk fills it.
+			m.messages[i].content = text
+		} else {
+			// Streaming tools send incremental chunks; append the raw delta
+			// so line boundaries are preserved inside the same card.
+			m.messages[i].content += strings.TrimPrefix(text, incomingHeader+"\n[output]\n")
+		}
+		return true
 	}
 	return false
 }
@@ -668,72 +726,25 @@ type metaPill struct {
 	accent bool
 }
 
-func (m model) renderMetaPills(parts []metaPill) string {
-	if len(parts) == 0 {
-		return ""
+func (m model) inputModeView() string {
+	if strings.HasPrefix(strings.TrimSpace(m.input.Value()), "!") {
+		return m.styles.InputModeCommand.Render(iconCommand)
 	}
-
-	plain := make([]string, len(parts))
-	for i, part := range parts {
-		plain[i] = part.text
+	if m.yolo {
+		return m.styles.InputModeYolo.Render("Y")
 	}
-
-	width := max(1, m.width-2)
-	visible := len(plain)
-	for visible > 1 && lipgloss.Width(strings.Join(plain[:visible], " ")) > width {
-		visible--
-	}
-	if lipgloss.Width(strings.Join(plain[:visible], " ")) > width {
-		return m.styles.PillAccent.Render(fitLine(plain[0], width))
-	}
-
-	rendered := make([]string, 0, visible)
-	for i := 0; i < visible; i++ {
-		style := m.styles.Pill
-		if parts[i].accent {
-			style = m.styles.PillAccent
-		}
-		rendered = append(rendered, style.Render(parts[i].text))
-	}
-	return strings.Join(rendered, " ")
+	return m.styles.InputModeReadonly.Render("R")
 }
 
-func (m model) modelMetaView() string {
-	mode := "read-only tools"
-	if m.yolo {
-		mode = "yolo tools"
-	}
-
-	modelName := m.modelName
-	if m.models != nil {
-		if currentModel := m.models.CurrentModel(); currentModel != "" {
-			modelName = currentModel
-		}
-	}
-
-	parts := []metaPill{{text: iconModel + " " + modelName, accent: true}}
-	if m.models != nil {
-		if provider := m.models.CurrentProvider(); provider != "" {
-			parts = append(parts, metaPill{text: "provider " + provider})
-		}
-		if currentThinking := m.models.CurrentThinking(); currentThinking != "" && currentThinking != "unset" {
-			parts = append(parts, metaPill{text: "thinking " + currentThinking})
-		}
-		if sessionName := m.models.CurrentSessionName(); sessionName != "" {
-			parts = append(parts, metaPill{text: iconSession + " " + sessionName})
-		}
-	}
-	if m.contextUsage != "" {
-		parts = append(parts, metaPill{text: iconContext + " " + m.contextUsage})
-	}
-	parts = append(parts, metaPill{text: mode}, metaPill{text: "theme " + string(m.theme.Mode)})
-
-	return m.styles.ModelMeta.Width(max(1, m.width)).Render(m.renderMetaPills(parts))
+func (m model) inputModeWidth() int {
+	return lipgloss.Width(m.inputModeView())
 }
 
 func (m model) inputView() string {
 	palette := m.commandPaletteView()
-	box := m.styles.InputBox.Width(max(20, m.width-2)).Render(m.input.View())
+	gap := lipgloss.NewStyle().Background(m.styles.InputBox.GetBackground()).Render(" ")
+	content := lipgloss.JoinHorizontal(lipgloss.Top, m.inputModeView(), gap, m.input.View())
+	box := m.styles.InputBox.Width(max(20, m.width)).Render(content)
 	if palette != "" {
 		return lipgloss.JoinVertical(lipgloss.Left, palette, box)
 	}
@@ -751,13 +762,85 @@ func (m model) activityView() string {
 	line := m.styles.StatusInfo.Render(iconWorking) + " Working " + m.spinner.View() + " " + status + " · esc cancels"
 	return m.styles.Activity.Render(fitLine(line, max(1, m.width-2)))
 }
-
 func (m model) footerView() string {
-	hint := "enter send · ctrl+j newline · alt+enter newline · ctrl+p/ctrl+n history · / commands"
+	hint := ""
 	if m.running {
 		hint = "esc cancel · ctrl+c quit"
 	}
-	return m.styles.Footer.Width(max(1, m.width)).Render(fitLine(hint, max(1, m.width-2)))
+	if m.showCommandPalette() {
+		if sug, ok := m.selectedCommandSuggestion(); ok && sug.executable {
+			hint = "↑/↓ select · enter apply · tab fill"
+		}
+	}
+
+	inner := max(1, m.width-2) // Footer padding(0,1)
+	pills := m.modelMetaPillsClamped(inner / 3)
+	pillsW := lipgloss.Width(pills)
+
+	var line string
+	if pills != "" {
+		remain := max(1, inner-pillsW-1)
+		line = pills + " " + fitLine(hint, remain)
+	} else {
+		line = fitLine(hint, inner)
+	}
+	return m.styles.Footer.Width(max(1, m.width)).Render(line)
+}
+
+func (m model) modelMetaPillsClamped(maxW int) string {
+	modelName := m.modelName
+	if m.models != nil {
+		if currentModel := m.models.CurrentModel(); currentModel != "" {
+			modelName = currentModel
+		}
+	}
+
+	parts := []metaPill{{text: iconModel + " " + modelName, accent: true}}
+	if m.models != nil {
+		if provider := m.models.CurrentProvider(); provider != "" {
+			parts = append(parts, metaPill{text: provider})
+		}
+		if currentThinking := m.models.CurrentThinking(); currentThinking != "" && currentThinking != "unset" {
+			parts = append(parts, metaPill{text: currentThinking})
+		}
+		if sessionName := m.models.CurrentSessionName(); sessionName != "" {
+			parts = append(parts, metaPill{text: iconSession + " " + sessionName})
+		}
+	}
+	if m.contextUsage != "" {
+		parts = append(parts, metaPill{text: iconContext + " " + m.contextUsage})
+	}
+
+	return m.renderMetaPillsClamped(parts, maxW)
+}
+
+func (m model) renderMetaPillsClamped(parts []metaPill, width int) string {
+	if len(parts) == 0 {
+		return ""
+	}
+
+	plain := make([]string, len(parts))
+	for i, part := range parts {
+		plain[i] = part.text
+	}
+
+	visible := len(plain)
+	for visible > 1 && lipgloss.Width(strings.Join(plain[:visible], " ")) > width {
+		visible--
+	}
+	if visible > 0 && lipgloss.Width(strings.Join(plain[:visible], " ")) > width {
+		return ""
+	}
+
+	rendered := make([]string, 0, visible)
+	for i := 0; i < visible; i++ {
+		style := m.styles.Pill
+		if parts[i].accent {
+			style = m.styles.PillAccent
+		}
+		rendered = append(rendered, style.Render(parts[i].text))
+	}
+	return strings.Join(rendered, " ")
 }
 
 type eventWriter struct {
@@ -801,6 +884,7 @@ func isViewportKey(key string) bool {
 }
 
 func (m *model) afterInputChanged() {
+	m.commandCursor = 0
 	m.syncInputHeight()
 	if m.width > 0 && m.height > 0 {
 		m.resize()
@@ -882,18 +966,37 @@ func (m model) inputBlockHeight() int {
 }
 
 func (m model) commandPaletteHeight() int {
-	matches := commandMatches(commandPaletteQuery(m.input.Value()))
-	count := len(matches)
-	if count > maxCommandPaletteItems {
-		count = maxCommandPaletteItems
+	sugs := m.commandSuggestions()
+	total := len(sugs)
+	if !m.showCommandPalette() {
+		return 0
 	}
+	count := min(total, maxCommandPaletteItems)
 	if count == 0 {
 		count = 1
 	}
-	return 1 + count + m.styles.Palette.GetVerticalFrameSize()
+	scrollUp := 0
+	scrollDown := 0
+	if total > maxCommandPaletteItems {
+		visible := min(total, maxCommandPaletteItems)
+		scroll := max(0, m.commandCursor-visible+1)
+		if m.commandCursor < scroll {
+			scroll = m.commandCursor
+		}
+		if scroll > 0 {
+			scrollUp = 1
+		}
+		if scroll+visible < total {
+			scrollDown = 1
+		}
+	}
+	return 1 + scrollUp + count + scrollDown + m.styles.Palette.GetVerticalFrameSize()
 }
 
 func (m model) showCommandPalette() bool {
+	if len(m.commandSuggestions()) > 0 {
+		return true
+	}
 	value := strings.TrimSpace(m.input.Value())
 	return strings.HasPrefix(value, "/") && !strings.Contains(value, " ")
 }
@@ -902,18 +1005,42 @@ func (m model) commandPaletteView() string {
 	if !m.showCommandPalette() {
 		return ""
 	}
-	matches := commandMatches(commandPaletteQuery(m.input.Value()))
-	lines := []string{m.styles.Header.Render(iconCommand + " Commands")}
-	if len(matches) == 0 {
-		lines = append(lines, m.styles.Muted.Render("No matching commands"))
-	} else {
-		for i, match := range matches {
-			if i >= maxCommandPaletteItems {
-				break
-			}
-			lines = append(lines, fmt.Sprintf("%s %s", m.styles.PaletteMatch.Render(match.command.Name), m.styles.Muted.Render(match.command.Description)))
+	sugs := m.commandSuggestions()
+	if len(sugs) > 0 {
+		header := iconCommand + " Commands"
+		if sugs[0].executable {
+			header = iconCommand + " Model candidates"
 		}
+		lines := []string{m.styles.Header.Render(header)}
+		m.clampCommandCursor()
+		total := len(sugs)
+		visible := min(total, maxCommandPaletteItems)
+		scroll := max(0, m.commandCursor-visible+1)
+		if m.commandCursor < scroll {
+			scroll = m.commandCursor
+		}
+		if scroll > 0 {
+			lines = append(lines, m.styles.Muted.Render("  ↑ more"))
+		}
+		for i := scroll; i < scroll+visible && i < total; i++ {
+			s := sugs[i]
+			prefix := "  "
+			if i == m.commandCursor {
+				prefix = "▸ "
+			}
+			lines = append(lines, fmt.Sprintf("%s%s %s", prefix, m.styles.PaletteMatch.Render(s.label), m.styles.Muted.Render(s.description)))
+		}
+		if scroll+visible < total {
+			lines = append(lines, m.styles.Muted.Render("  ↓ more"))
+		}
+		return m.styles.Palette.Width(max(20, m.width-2)).Render(strings.Join(lines, "\n"))
 	}
+	value := strings.TrimSpace(m.input.Value())
+	if strings.Contains(value, " ") {
+		return ""
+	}
+	lines := []string{m.styles.Header.Render(iconCommand + " Commands")}
+	lines = append(lines, m.styles.Muted.Render("No matching commands"))
 	return m.styles.Palette.Width(max(20, m.width-2)).Render(strings.Join(lines, "\n"))
 }
 
@@ -991,6 +1118,162 @@ func subsequenceScore(candidate string, query string, base int) (int, bool) {
 	return base + gaps, true
 }
 
+func (m model) commandSuggestions() []commandSuggestion {
+	input := m.input.Value()
+	if sugs := m.modelCommandSuggestions(input); len(sugs) > 0 {
+		return sugs
+	}
+	value := strings.TrimSpace(input)
+	if strings.HasPrefix(value, "/") && !strings.Contains(value, " ") {
+		return slashCommandSuggestions(commandPaletteQuery(input))
+	}
+	return nil
+}
+
+func slashCommandSuggestions(query string) []commandSuggestion {
+	matches := commandMatches(query)
+	sugs := make([]commandSuggestion, 0, len(matches))
+	for _, m := range matches {
+		sugs = append(sugs, commandSuggestion{
+			label:       m.command.Name,
+			description: m.command.Description,
+			value:       m.command.Name,
+		})
+	}
+	return sugs
+}
+
+func splitModelCommandTail(input string) (query string, ok bool) {
+	fields := strings.Fields(input)
+	if len(fields) == 0 || fields[0] != "/model" {
+		return "", false
+	}
+	hasTrailingSpace := strings.TrimRight(input, " \t") != input
+	switch len(fields) {
+	case 1:
+		if !hasTrailingSpace {
+			return "", false
+		}
+		return "", true
+	case 2:
+		if fields[1] == "provider" || fields[1] == "thinking" {
+			return "", false
+		}
+		if fields[1] == "model" {
+			if !hasTrailingSpace {
+				return "", false
+			}
+			return "", true
+		}
+		if hasTrailingSpace {
+			return fields[1], true
+		}
+		return fields[1], true
+	case 3:
+		if fields[1] != "model" {
+			return "", false
+		}
+		if hasTrailingSpace {
+			return fields[2], true
+		}
+		return fields[2], true
+	default:
+		return "", false
+	}
+}
+
+func (m model) modelCommandSuggestions(input string) []commandSuggestion {
+	if m.models == nil {
+		return nil
+	}
+	query, ok := splitModelCommandTail(input)
+	if !ok {
+		return nil
+	}
+	available := m.models.AvailableModels()
+	if len(available) == 0 {
+		return nil
+	}
+	current := m.models.CurrentModel()
+	candidates := make([]commandSuggestion, 0, len(available))
+	for _, id := range available {
+		desc := "switch model"
+		if id == current {
+			desc = "current model"
+		}
+		candidates = append(candidates, commandSuggestion{
+			label:       id,
+			description: desc,
+			value:       "/model model " + id,
+			executable:  true,
+		})
+	}
+	return scoredCommandSuggestions(candidates, query)
+}
+
+func scoredCommandSuggestions(candidates []commandSuggestion, query string) []commandSuggestion {
+	if query == "" {
+		return candidates
+	}
+	type scored struct {
+		sug   commandSuggestion
+		score int
+		idx   int
+	}
+	var kept []scored
+	for i, c := range candidates {
+		if s, ok := scoreCommandSuggestion(c.label, query); ok {
+			kept = append(kept, scored{sug: c, score: s, idx: i})
+		}
+	}
+	sort.Slice(kept, func(i, j int) bool {
+		if kept[i].score != kept[j].score {
+			return kept[i].score < kept[j].score
+		}
+		return kept[i].idx < kept[j].idx
+	})
+	out := make([]commandSuggestion, 0, len(kept))
+	for _, s := range kept {
+		out = append(out, s.sug)
+	}
+	return out
+}
+
+func scoreCommandSuggestion(label string, query string) (int, bool) {
+	query = strings.ToLower(strings.TrimSpace(query))
+	name := strings.ToLower(label)
+	if query == "" || query == name {
+		return 0, true
+	}
+	if strings.HasPrefix(name, query) {
+		return 100 + len(name) - len(query), true
+	}
+	return subsequenceScore(name, query, 200)
+}
+
+func (m model) selectedCommandSuggestion() (commandSuggestion, bool) {
+	sugs := m.commandSuggestions()
+	m.clampCommandCursor()
+	if m.commandCursor >= 0 && m.commandCursor < len(sugs) {
+		return sugs[m.commandCursor], true
+	}
+	return commandSuggestion{}, false
+}
+
+func (m *model) clampCommandCursor() {
+	sugs := m.commandSuggestions()
+	if len(sugs) == 0 {
+		m.commandCursor = 0
+		return
+	}
+	if m.commandCursor < 0 {
+		m.commandCursor = 0
+	}
+	if m.commandCursor >= len(sugs) {
+		m.commandCursor = len(sugs) - 1
+	}
+}
+
 func (m model) slashSuggestionCount() int {
 	return m.commandPaletteHeight()
 }
@@ -1007,11 +1290,7 @@ func (m model) modelInfo() string {
 	if m.models != nil {
 		return m.models.Info()
 	}
-	mode := "read-only tools"
-	if m.yolo {
-		mode = "yolo tools"
-	}
-	return fmt.Sprintf("model: %s\nmode: %s\ntheme: %s", m.modelName, mode, m.theme.Mode)
+	return m.modelName
 }
 
 func (m *model) refreshContextUsage() {
@@ -1090,6 +1369,9 @@ func fitLine(text string, width int) string {
 }
 
 func (m *model) runLocalCommand(input string) (tea.Model, tea.Cmd) {
+	if strings.HasPrefix(input, "/theme") {
+		return m.handleThemeCommand(input)
+	}
 	output, err := m.handleLocalCommand(input)
 	if err == errOpenPicker {
 		m.openSessionPicker()
@@ -1114,7 +1396,60 @@ func (m *model) runLocalCommand(input string) (tea.Model, tea.Cmd) {
 	return *m, nil
 }
 
+func (m *model) handleThemeCommand(input string) (tea.Model, tea.Cmd) {
+	fields := strings.Fields(input)
+	if len(fields) < 2 {
+		m.input.SetValue("")
+		m.afterInputChanged()
+		m.messages = append(m.messages, message{
+			role: roleSystem, content: "theme: " + string(m.theme.Mode) + "\nusage: /theme light|dark",
+		})
+		m.refreshViewport(true)
+		return *m, nil
+	}
+	mode, err := ParseThemeMode(fields[1])
+	if err != nil {
+		m.input.SetValue("")
+		m.afterInputChanged()
+		m.messages = append(m.messages, message{role: roleError, content: err.Error()})
+		m.refreshViewport(true)
+		return *m, nil
+	}
+	m.theme = themeForMode(mode)
+	m.styles = m.theme.styles()
+	applyTextareaTheme(&m.input, m.theme)
+	m.input.SetValue("")
+	m.afterInputChanged()
+	m.messages = append(m.messages, message{role: roleSystem, content: "theme: " + string(mode)})
+	m.refreshViewport(true)
+	return *m, nil
+}
+
+func (m model) handleBangCommand(input string) (string, error) {
+	command := strings.TrimSpace(strings.TrimPrefix(strings.TrimSpace(input), "!"))
+	if command == "" {
+		return "", fmt.Errorf("usage: ! <command>")
+	}
+	if !m.yolo {
+		return "", fmt.Errorf("shell command input requires --yolo")
+	}
+	out, err := m.runShellCommand(command)
+	if err != nil {
+		if strings.TrimSpace(out) != "" {
+			return "", fmt.Errorf("%s\n%w", strings.TrimRight(out, "\n"), err)
+		}
+		return "", err
+	}
+	if strings.TrimSpace(out) == "" {
+		return fmt.Sprintf("$ %s\n(no output)", command), nil
+	}
+	return out, nil
+}
+
 func (m model) handleLocalCommand(input string) (string, error) {
+	if strings.HasPrefix(strings.TrimSpace(input), "!") {
+		return m.handleBangCommand(input)
+	}
 	fields := strings.Fields(input)
 	if len(fields) == 0 {
 		return "", fmt.Errorf("empty command")
@@ -1310,6 +1645,41 @@ func (m *model) updatePicker(msg tea.KeyMsg) (tea.Model, tea.Cmd) {
 	return *m, nil
 }
 
+func (m *model) updateCommandSuggestions(msg tea.KeyMsg) (bool, tea.Model, tea.Cmd) {
+	if m.running || len(m.commandSuggestions()) == 0 {
+		return false, *m, nil
+	}
+	m.clampCommandCursor()
+	switch msg.String() {
+	case "up":
+		if m.commandCursor > 0 {
+			m.commandCursor--
+		}
+		return true, *m, nil
+	case "down":
+		if m.commandCursor < len(m.commandSuggestions())-1 {
+			m.commandCursor++
+		}
+		return true, *m, nil
+	case "tab":
+		if sug, ok := m.selectedCommandSuggestion(); ok {
+			m.input.SetValue(sug.value)
+			m.afterInputChanged()
+		}
+		return true, *m, nil
+	case "enter":
+		if sug, ok := m.selectedCommandSuggestion(); ok && sug.executable {
+			m.rememberInput(sug.value)
+			m.input.SetValue(sug.value)
+			m.afterInputChanged()
+			next, cmd := m.runLocalCommand(sug.value)
+			return true, next, cmd
+		}
+		return false, *m, nil
+	}
+	return false, *m, nil
+}
+
 func (m model) pickerView() string {
 	if !m.picking || len(m.pickerItems) == 0 {
 		return ""

internal/tui/app_test.go

@@ -21,7 +21,7 @@ func TestModelRendersChatShell(t *testing.T) {
 	m = next.(model)
 	rendered := m.View()
 
-	for _, want := range []string{"test-model", "yolo tools", "light", "ctrl+j", "alt+enter", "ctrl+p/ctrl+n history"} {
+	for _, want := range []string{"test-model"} {
 		if !strings.Contains(rendered, want) {
 			t.Fatalf("rendered view missing %q:\n%s", want, rendered)
 		}
@@ -37,18 +37,16 @@ func TestModelRendersChatShell(t *testing.T) {
 	}
 }
 
-func TestInputTextDoesNotPaintBackground(t *testing.T) {
+func TestInputTextBackgroundMatchesInputBox(t *testing.T) {
 	m := newModel(context.Background(), nil, Options{ModelName: "test-model"})
-	for name, color := range map[string]any{
-		"input box":           m.styles.InputBox.GetBackground(),
+	want := lightTheme().InputBg
+	for name, color := range map[string]lipgloss.TerminalColor{
 		"focused base":        m.input.FocusedStyle.Base.GetBackground(),
 		"focused cursor line": m.input.FocusedStyle.CursorLine.GetBackground(),
 		"blurred base":        m.input.BlurredStyle.Base.GetBackground(),
 		"blurred cursor line": m.input.BlurredStyle.CursorLine.GetBackground(),
 	} {
-		if _, ok := color.(lipgloss.NoColor); !ok {
-			t.Fatalf("%s should not paint input text background", name)
-		}
+		requireStyleColor(t, name, color, want)
 	}
 }
 
@@ -58,12 +56,12 @@ func TestInputBoxIsRoomier(t *testing.T) {
 	m = next.(model)
 
 	if got := m.styles.InputBox.GetVerticalFrameSize(); got != 2 {
-		t.Fatalf("input box vertical frame should stay compact, got %d", got)
+		t.Fatalf("input box vertical frame should be 2 with vertical padding, got %d", got)
 	}
-	if got := m.styles.InputBox.GetHorizontalFrameSize(); got < 6 {
-		t.Fatalf("input box horizontal frame should include roomier padding, got %d", got)
+	if got := m.styles.InputBox.GetHorizontalFrameSize(); got != 3 {
+		t.Fatalf("input box horizontal frame should be 3 (asymmetric padding), got %d", got)
 	}
-	if got, want := m.input.Width(), 100-m.styles.InputBox.GetHorizontalFrameSize(); got != want {
+	if got, want := m.input.Width(), 100-m.styles.InputBox.GetHorizontalFrameSize()-m.inputModeWidth()-1; got != want {
 		t.Fatalf("input content width = %d, want %d", got, want)
 	}
 	rendered := m.inputView()
@@ -78,13 +76,8 @@ func TestModelMetaRendersBelowInput(t *testing.T) {
 	next, _ := m.Update(tea.WindowSizeMsg{Width: 100, Height: 30})
 	rendered := next.(model).View()
 
-	inputIndex := strings.Index(rendered, "Message agentu")
-	modelIndex := strings.Index(rendered, "test-model")
-	if inputIndex < 0 || modelIndex < 0 {
-		t.Fatalf("rendered view missing input or model meta:\n%s", rendered)
-	}
-	if modelIndex < inputIndex {
-		t.Fatalf("model meta should render below input:\n%s", rendered)
+	if !strings.Contains(rendered, "test-model") {
+		t.Fatalf("rendered view missing %q:\n%s", "test-model", rendered)
 	}
 }
 
@@ -93,7 +86,7 @@ func TestModelMetaRendersContextUsage(t *testing.T) {
 	assistant.SetMessages([]llm.Message{{Role: llm.RoleUser, Content: strings.Repeat("x", 800)}})
 	m := newModel(context.Background(), assistant, Options{ModelName: "test-model"})
 	next, _ := m.Update(tea.WindowSizeMsg{Width: 120, Height: 30})
-	rendered := next.(model).modelMetaView()
+	rendered := next.(model).footerView()
 
 	for _, want := range []string{"ctx ~", "/1k", "%"} {
 		if !strings.Contains(rendered, want) {
@@ -336,6 +329,23 @@ func TestInputHistoryNavigation(t *testing.T) {
 	}
 }
 
+func TestUpArrowRecallsPreviousInput(t *testing.T) {
+	m := newModel(context.Background(), nil, Options{ModelName: "test-model"})
+	m.rememberInput("first")
+	m.rememberInput("second")
+
+	next, _ := m.Update(tea.KeyMsg{Type: tea.KeyUp})
+	m = next.(model)
+	if got := m.input.Value(); got != "second" {
+		t.Fatalf("after first up input = %q, want %q", got, "second")
+	}
+	next, _ = m.Update(tea.KeyMsg{Type: tea.KeyUp})
+	m = next.(model)
+	if got := m.input.Value(); got != "first" {
+		t.Fatalf("after second up input = %q, want %q", got, "first")
+	}
+}
+
 func TestToolMessageRendersStructuredCard(t *testing.T) {
 	m := newModel(context.Background(), nil, Options{ModelName: "test-model"})
 	m.messages = append(m.messages, message{role: roleTool, content: `file_read {"path":"README.md"}`})
@@ -385,6 +395,28 @@ func TestToolLogOutputUpdatesExistingToolCard(t *testing.T) {
 	}
 }
 
+func TestToolLogStreamingChunksAppendToSameCard(t *testing.T) {
+	m := newModel(context.Background(), nil, Options{ModelName: "test-model"})
+	m.messages = append(m.messages, message{role: roleTool, content: `shell_run echo hi`})
+
+	if !m.mergeToolLog("shell_run echo hi\n[output]\none\n") {
+		t.Fatal("expected first mergeToolLog to return true")
+	}
+	if !m.mergeToolLog("shell_run echo hi\n[output]\ntwo\n") {
+		t.Fatal("expected second mergeToolLog to return true")
+	}
+	if len(m.messages) != 1 {
+		t.Fatalf("messages count = %d, want 1", len(m.messages))
+	}
+	plain := stripANSI(m.renderMessages())
+	if !strings.Contains(plain, "one") || !strings.Contains(plain, "two") {
+		t.Fatalf("rendered output missing streamed lines:\n%s", plain)
+	}
+	if strings.Contains(plain, "onetwo") {
+		t.Fatalf("streamed lines were concatenated:\n%s", plain)
+	}
+}
+
 type tuiCompactProvider struct{}
 
 func (tuiCompactProvider) ChatStream(ctx context.Context, req llm.ChatRequest, emit func(llm.StreamEvent) error) error {
@@ -480,16 +512,60 @@ func runInDir(t *testing.T, dir string, name string, args ...string) {
 	}
 }
 
+func TestBangShellCommandRunsFromInput(t *testing.T) {
+	dir := t.TempDir()
+	m := newModel(context.Background(), nil, Options{ModelName: "test-model", Yolo: true, WorkingDir: dir})
+	next, _ := m.Update(tea.WindowSizeMsg{Width: 100, Height: 30})
+	m = next.(model)
+	m.input.SetValue("! printf hello")
+	next, _ = m.submit()
+	m = next.(model)
+	if m.input.Value() != "" {
+		t.Fatalf("input not cleared after bang command: %q", m.input.Value())
+	}
+	rendered := stripANSI(m.renderMessages())
+	if !strings.Contains(rendered, "$ printf hello") {
+		t.Fatalf("rendered output missing shell echo:\n%s", rendered)
+	}
+	if !strings.Contains(rendered, "hello") {
+		t.Fatalf("rendered output missing 'hello':\n%s", rendered)
+	}
+}
+
+func TestBangShellCommandRequiresYolo(t *testing.T) {
+	m := newModel(context.Background(), nil, Options{ModelName: "test-model", Yolo: false})
+	next, _ := m.Update(tea.WindowSizeMsg{Width: 100, Height: 30})
+	m = next.(model)
+	m.input.SetValue("! printf hello")
+	next, _ = m.submit()
+	m = next.(model)
+	rendered := stripANSI(m.renderMessages())
+	if !strings.Contains(rendered, "shell command input requires --yolo") {
+		t.Fatalf("rendered output missing yolo requirement:\n%s", rendered)
+	}
+}
+
+func TestBangShellCommandRejectsEmptyCommand(t *testing.T) {
+	m := newModel(context.Background(), nil, Options{ModelName: "test-model", Yolo: true})
+	next, _ := m.Update(tea.WindowSizeMsg{Width: 100, Height: 30})
+	m = next.(model)
+	m.input.SetValue("!")
+	next, _ = m.submit()
+	m = next.(model)
+	rendered := stripANSI(m.renderMessages())
+	if !strings.Contains(rendered, "usage: ! <command>") {
+		t.Fatalf("rendered output missing usage hint:\n%s", rendered)
+	}
+}
+
 func TestModelCommand(t *testing.T) {
 	m := newModel(context.Background(), nil, Options{ModelName: "test-model", Yolo: true})
 	m.input.SetValue("/model")
 
 	next, _ := m.submit()
 	rendered := next.(model).renderMessages()
-	for _, want := range []string{"model: test-model", "mode: yolo tools", "theme: light"} {
-		if !strings.Contains(rendered, want) {
-			t.Fatalf("model command output missing %q:\n%s", want, rendered)
-		}
+	if !strings.Contains(rendered, "test-model") {
+		t.Fatalf("model command output missing %q:\n%s", "test-model", rendered)
 	}
 }
 
@@ -508,11 +584,124 @@ func TestModelThinkingCommand(t *testing.T) {
 	}
 }
 
+func TestModelCommandPaletteShowsConfiguredModels(t *testing.T) {
+	manager := &fakeModelManager{model: "model-a", models: []string{"model-a", "model-b"}}
+	m := newModel(context.Background(), nil, Options{ModelName: "model-a", ModelManager: manager})
+	next, _ := m.Update(tea.WindowSizeMsg{Width: 100, Height: 30})
+	m = next.(model)
+	m.input.SetValue("/model ")
+	m.afterInputChanged()
+	plain := stripANSI(m.View())
+	for _, want := range []string{iconCommand + " Model candidates", "model-a", "current model", "model-b", "switch model"} {
+		if !strings.Contains(plain, want) {
+			t.Fatalf("rendered palette missing %q:\n%s", want, plain)
+		}
+	}
+}
+
+func TestModelCommandPaletteSelectsConfiguredModel(t *testing.T) {
+	manager := &fakeModelManager{model: "model-a", models: []string{"model-a", "model-b"}}
+	m := newModel(context.Background(), nil, Options{ModelName: "model-a", ModelManager: manager})
+	next, _ := m.Update(tea.WindowSizeMsg{Width: 100, Height: 30})
+	m = next.(model)
+	m.input.SetValue("/model ")
+	m.afterInputChanged()
+	next, _ = m.Update(tea.KeyMsg{Type: tea.KeyDown})
+	m = next.(model)
+	next, _ = m.Update(tea.KeyMsg{Type: tea.KeyEnter})
+	m = next.(model)
+	if manager.model != "model-b" {
+		t.Fatalf("model = %q", manager.model)
+	}
+	if got := m.input.Value(); got != "" {
+		t.Fatalf("input = %q", got)
+	}
+	rendered := stripANSI(m.renderMessages())
+	if !strings.Contains(rendered, "model: model-b") {
+		t.Fatalf("rendered output missing model switch:\n%s", rendered)
+	}
+}
+
+func TestModelCommandPaletteFiltersBareModelPartial(t *testing.T) {
+	manager := &fakeModelManager{model: "claude-sonnet", models: []string{"claude-sonnet", "deepseek-v4-flash"}}
+	m := newModel(context.Background(), nil, Options{ModelName: "claude-sonnet", ModelManager: manager})
+	next, _ := m.Update(tea.WindowSizeMsg{Width: 100, Height: 30})
+	m = next.(model)
+	m.input.SetValue("/model deep")
+	m.afterInputChanged()
+	palette := stripANSI(m.commandPaletteView())
+	if !strings.Contains(palette, "deepseek-v4-flash") {
+		t.Fatalf("expected deepseek-v4-flash in palette:\n%s", palette)
+	}
+	if strings.Contains(palette, "claude-sonnet") {
+		t.Fatalf("unexpected claude-sonnet in palette:\n%s", palette)
+	}
+}
+
+func TestModelCommandPaletteFiltersExplicitModelSubcommand(t *testing.T) {
+	manager := &fakeModelManager{model: "claude-sonnet", models: []string{"claude-sonnet", "deepseek-v4-flash"}}
+	m := newModel(context.Background(), nil, Options{ModelName: "claude-sonnet", ModelManager: manager})
+	next, _ := m.Update(tea.WindowSizeMsg{Width: 100, Height: 30})
+	m = next.(model)
+	m.input.SetValue("/model model deep")
+	m.afterInputChanged()
+	palette := stripANSI(m.commandPaletteView())
+	if !strings.Contains(palette, "deepseek-v4-flash") {
+		t.Fatalf("expected deepseek-v4-flash in palette:\n%s", palette)
+	}
+	if strings.Contains(palette, "claude-sonnet") {
+		t.Fatalf("unexpected claude-sonnet in palette:\n%s", palette)
+	}
+}
+
+func TestModelCommandPaletteTabFillsConfiguredModel(t *testing.T) {
+	manager := &fakeModelManager{model: "claude-sonnet", models: []string{"claude-sonnet", "deepseek-v4-flash"}}
+	m := newModel(context.Background(), nil, Options{ModelName: "claude-sonnet", ModelManager: manager})
+	next, _ := m.Update(tea.WindowSizeMsg{Width: 100, Height: 30})
+	m = next.(model)
+	m.input.SetValue("/model deep")
+	m.afterInputChanged()
+	next, _ = m.Update(tea.KeyMsg{Type: tea.KeyTab})
+	m = next.(model)
+	if got := m.input.Value(); got != "/model model deepseek-v4-flash" {
+		t.Fatalf("input = %q", got)
+	}
+	if manager.model != "claude-sonnet" {
+		t.Fatalf("model should not change on tab: %q", manager.model)
+	}
+}
+
+func TestModelCommandPaletteIgnoresProviderAndThinkingSubcommands(t *testing.T) {
+	manager := &fakeModelManager{model: "model-a", models: []string{"model-a", "model-b"}}
+	m := newModel(context.Background(), nil, Options{ModelName: "model-a", ModelManager: manager})
+	next, _ := m.Update(tea.WindowSizeMsg{Width: 100, Height: 30})
+	m = next.(model)
+
+	m.input.SetValue("/model provider ")
+	m.afterInputChanged()
+	if sugs := m.commandSuggestions(); len(sugs) != 0 {
+		t.Fatalf("provider suggestions should be empty, got %d", len(sugs))
+	}
+	if view := m.commandPaletteView(); view != "" {
+		t.Fatalf("provider palette should be empty:\n%s", view)
+	}
+
+	m.input.SetValue("/model thinking ")
+	m.afterInputChanged()
+	if sugs := m.commandSuggestions(); len(sugs) != 0 {
+		t.Fatalf("thinking suggestions should be empty, got %d", len(sugs))
+	}
+	if view := m.commandPaletteView(); view != "" {
+		t.Fatalf("thinking palette should be empty:\n%s", view)
+	}
+}
+
 type fakeModelManager struct {
 	provider    string
 	model       string
 	thinking    string
 	sessionName string
+	models      []string
 	saved       bool
 }
 
@@ -583,6 +772,16 @@ func (f *fakeModelManager) NewSession() (string, error) {
 	return "New session.", nil
 }
 
+func (f *fakeModelManager) AvailableModels() []string {
+	if len(f.models) > 0 {
+		return append([]string(nil), f.models...)
+	}
+	if f.model != "" {
+		return []string{f.model}
+	}
+	return nil
+}
+
 func TestParseThemeMode(t *testing.T) {
 	for input, want := range map[string]ThemeMode{
 		"":      ThemeLight,
@@ -604,6 +803,130 @@ func TestParseThemeMode(t *testing.T) {
 	}
 }
 
+func requireStyleColor(t *testing.T, name string, got lipgloss.TerminalColor, want string) {
+	t.Helper()
+	color, ok := got.(lipgloss.Color)
+	if !ok {
+		t.Fatalf("%s = %T, want lipgloss.Color", name, got)
+	}
+	if string(color) != want {
+		t.Fatalf("%s = %q, want %q", name, string(color), want)
+	}
+}
+
+func requireNoStyleColor(t *testing.T, name string, got lipgloss.TerminalColor) {
+	t.Helper()
+	if _, ok := got.(lipgloss.NoColor); !ok {
+		t.Fatalf("%s should not paint a color, got %T", name, got)
+	}
+}
+
+func TestDarkThemePaintsAppBackground(t *testing.T) {
+	m := newModel(context.Background(), nil, Options{ModelName: "test-model", ThemeMode: ThemeDark})
+	next, _ := m.Update(tea.WindowSizeMsg{Width: 100, Height: 30})
+	m = next.(model)
+	rendered := stripANSI(m.View())
+	_ = rendered
+	if m.theme.Mode != ThemeDark {
+		t.Fatalf("theme.Mode = %q, want %q", m.theme.Mode, ThemeDark)
+	}
+	want := darkTheme()
+	requireStyleColor(t, "App.Background", m.styles.App.GetBackground(), want.Background)
+	requireStyleColor(t, "App.Foreground", m.styles.App.GetForeground(), want.Text)
+}
+
+func TestDarkThemeKeepsComponentBackgroundsLightweight(t *testing.T) {
+	m := newModel(context.Background(), nil, Options{ModelName: "test-model", ThemeMode: ThemeDark})
+	next, _ := m.Update(tea.WindowSizeMsg{Width: 100, Height: 30})
+	m = next.(model)
+	requireNoStyleColor(t, "Viewport.Background", m.styles.Viewport.GetBackground())
+	requireStyleColor(t, "InputBox.Background", m.styles.InputBox.GetBackground(), darkTheme().InputBg)
+	requireNoStyleColor(t, "Activity.Background", m.styles.Activity.GetBackground())
+	m.running = true
+	m.status = "thinking..."
+	m.resize()
+	if backgroundPattern.MatchString(m.activityView()) {
+		t.Fatalf("activity view should not paint a background:\n%q", m.activityView())
+	}
+}
+
+func TestDarkThemeStylesUseDarkPalette(t *testing.T) {
+	m := newModel(context.Background(), nil, Options{ModelName: "test-model", ThemeMode: ThemeDark})
+	want := darkTheme()
+	requireStyleColor(t, "Header.Foreground", m.styles.Header.GetForeground(), want.Title)
+	requireStyleColor(t, "Muted.Foreground", m.styles.Muted.GetForeground(), want.Muted)
+	requireStyleColor(t, "Pill.Background", m.styles.Pill.GetBackground(), want.SurfaceAlt)
+	requireStyleColor(t, "input.FocusedStyle.Base.Foreground", m.input.FocusedStyle.Base.GetForeground(), want.Text)
+	requireStyleColor(t, "input.FocusedStyle.Placeholder.Foreground", m.input.FocusedStyle.Placeholder.GetForeground(), want.Muted)
+	requireStyleColor(t, "InputModeYolo.Foreground", m.styles.InputModeYolo.GetForeground(), want.Error)
+	requireStyleColor(t, "InputModeCommand.Foreground", m.styles.InputModeCommand.GetForeground(), want.User)
+}
+
+func TestInputModeSymbolReflectsYoloAndBangCommand(t *testing.T) {
+	m := newModel(context.Background(), nil, Options{ModelName: "test-model", Yolo: true})
+	next, _ := m.Update(tea.WindowSizeMsg{Width: 100, Height: 30})
+	m = next.(model)
+
+	rendered := stripANSI(m.inputView())
+	if !strings.Contains(rendered, "Y") {
+		t.Fatalf("YOLO input view should contain Y symbol:\n%s", rendered)
+	}
+
+	m.input.SetValue("! pwd")
+	m.afterInputChanged()
+	rendered = stripANSI(m.inputView())
+	if !strings.Contains(rendered, iconCommand) {
+		t.Fatalf("bang input view should contain command icon:\n%s", rendered)
+	}
+	if strings.Contains(rendered, "Y") {
+		t.Fatalf("bang input view should not contain Y symbol:\n%s", rendered)
+	}
+}
+
+func TestInputModeYoloUsesWarningColor(t *testing.T) {
+	m := newModel(context.Background(), nil, Options{ModelName: "test-model", Yolo: true})
+	requireStyleColor(t, "InputModeYolo.Foreground", m.styles.InputModeYolo.GetForeground(), lightTheme().Error)
+}
+
+func TestThemeSlashCommandSwitchesAtRuntime(t *testing.T) {
+	m := newModel(context.Background(), nil, Options{ModelName: "test-model"})
+	if m.theme.Mode != ThemeLight {
+		t.Fatalf("initial theme = %q, want light", m.theme.Mode)
+	}
+	requireNoStyleColor(t, "initial App.Background", m.styles.App.GetBackground())
+
+	// Switch to dark
+	m.input.SetValue("/theme dark")
+	next, _ := m.submit()
+	m = next.(model)
+	if m.theme.Mode != ThemeDark {
+		t.Fatalf("after /theme dark: Mode = %q, want dark", m.theme.Mode)
+	}
+	requireStyleColor(t, "dark App.Background", m.styles.App.GetBackground(), darkTheme().Background)
+	last := m.messages[len(m.messages)-1]
+	if !strings.Contains(last.content, "theme: dark") {
+		t.Fatalf("expected confirmation message, got %q", last.content)
+	}
+
+	// Switch back to light
+	m.input.SetValue("/theme light")
+	next, _ = m.submit()
+	m = next.(model)
+	if m.theme.Mode != ThemeLight {
+		t.Fatalf("after /theme light: Mode = %q, want light", m.theme.Mode)
+	}
+	requireNoStyleColor(t, "restored App.Background", m.styles.App.GetBackground())
+
+	// No arg shows usage
+	m.input.SetValue("/theme")
+	next, _ = m.submit()
+	m = next.(model)
+	last = m.messages[len(m.messages)-1]
+	if !strings.Contains(last.content, "usage: /theme") {
+		t.Fatalf("expected usage hint, got %q", last.content)
+	}
+}
+
 func TestCleanToolLog(t *testing.T) {
 	got := cleanToolLog("\n[tool] shell_run {\"command\":\"pwd\"}\n")
 	want := `shell_run {"command":"pwd"}`

internal/tui/theme.go

@@ -43,6 +43,7 @@ type theme struct {
 	Background string
 	Surface    string
 	SurfaceAlt string
+	InputBg    string
 	Border     string
 	Text       string
 	Muted      string
@@ -56,35 +57,37 @@ type theme struct {
 }
 
 type styles struct {
-	App           lipgloss.Style
-	Muted         lipgloss.Style
-	Activity      lipgloss.Style
-	Header        lipgloss.Style
-	ModelMeta     lipgloss.Style
-	Footer        lipgloss.Style
-	Viewport      lipgloss.Style
-	InputBox      lipgloss.Style
-	CommandHint   lipgloss.Style
-	StatusReady   lipgloss.Style
-	StatusWarn    lipgloss.Style
-	StatusError   lipgloss.Style
-	StatusInfo    lipgloss.Style
-	Pill          lipgloss.Style
-	PillAccent    lipgloss.Style
-	Palette       lipgloss.Style
-	PaletteMatch  lipgloss.Style
-	ToolName      lipgloss.Style
-	ToolKey       lipgloss.Style
-	ToolValue     lipgloss.Style
-	UserLabel     lipgloss.Style
-	ToolLabel     lipgloss.Style
-	ErrorLabel    lipgloss.Style
-	SystemLabel   lipgloss.Style
-	UserMessage   lipgloss.Style
-	AssistantMsg  lipgloss.Style
-	ToolMessage   lipgloss.Style
-	ErrorMessage  lipgloss.Style
-	SystemMessage lipgloss.Style
+	App               lipgloss.Style
+	Muted             lipgloss.Style
+	Activity          lipgloss.Style
+	Header            lipgloss.Style
+	Footer            lipgloss.Style
+	Viewport          lipgloss.Style
+	InputBox          lipgloss.Style
+	CommandHint       lipgloss.Style
+	InputModeReadonly lipgloss.Style
+	InputModeYolo     lipgloss.Style
+	InputModeCommand  lipgloss.Style
+	StatusReady       lipgloss.Style
+	StatusWarn        lipgloss.Style
+	StatusError       lipgloss.Style
+	StatusInfo        lipgloss.Style
+	Pill              lipgloss.Style
+	PillAccent        lipgloss.Style
+	Palette           lipgloss.Style
+	PaletteMatch      lipgloss.Style
+	ToolName          lipgloss.Style
+	ToolKey           lipgloss.Style
+	ToolValue         lipgloss.Style
+	UserLabel         lipgloss.Style
+	ToolLabel         lipgloss.Style
+	ErrorLabel        lipgloss.Style
+	SystemLabel       lipgloss.Style
+	UserMessage       lipgloss.Style
+	AssistantMsg      lipgloss.Style
+	ToolMessage       lipgloss.Style
+	ErrorMessage      lipgloss.Style
+	SystemMessage     lipgloss.Style
 }
 
 func themeForMode(mode ThemeMode) theme {
@@ -98,8 +101,8 @@ func lightTheme() theme {
 	return theme{
 		Mode:       ThemeLight,
 		Background: "#F8FAFC",
-		Surface:    "#FFFFFF",
 		SurfaceAlt: "#EEF2F7",
+		InputBg:    "#E4E4E4",
 		Border:     "#CBD5E1",
 		Text:       "#111827",
 		Muted:      "#64748B",
@@ -116,8 +119,8 @@ func darkTheme() theme {
 	return theme{
 		Mode:       ThemeDark,
 		Background: "#111827",
-		Surface:    "#1F2937",
 		SurfaceAlt: "#0F172A",
+		InputBg:    "#383838",
 		Border:     "#475569",
 		Text:       "#E5E7EB",
 		Muted:      "#94A3B8",
@@ -130,10 +133,17 @@ func darkTheme() theme {
 	}
 }
 
+func appStyle(t theme) lipgloss.Style {
+	style := lipgloss.NewStyle().Foreground(lipgloss.Color(t.Text))
+	if t.Mode == ThemeDark {
+		style = style.Background(lipgloss.Color(t.Background))
+	}
+	return style
+}
+
 func (t theme) styles() styles {
 	return styles{
-		App: lipgloss.NewStyle().
-			Foreground(lipgloss.Color(t.Text)),
+		App: appStyle(t),
 
 		Muted: lipgloss.NewStyle().
 			Foreground(lipgloss.Color(t.Muted)),
@@ -146,10 +156,6 @@ func (t theme) styles() styles {
 			Bold(true).
 			Foreground(lipgloss.Color(t.Title)),
 
-		ModelMeta: lipgloss.NewStyle().
-			Foreground(lipgloss.Color(t.Muted)).
-			Padding(0, 1),
-
 		Footer: lipgloss.NewStyle().
 			Foreground(lipgloss.Color(t.Muted)).
 			Padding(0, 1),
@@ -159,14 +165,31 @@ func (t theme) styles() styles {
 
 		InputBox: lipgloss.NewStyle().
 			Foreground(lipgloss.Color(t.Text)).
-			Border(lipgloss.ThickBorder()).
-			BorderForeground(lipgloss.Color(t.Border)).
-			Padding(0, 2),
+			Background(lipgloss.Color(t.InputBg)).
+			Padding(1, 2, 1, 1),
 
 		CommandHint: lipgloss.NewStyle().
 			Foreground(lipgloss.Color(t.Muted)).
 			Padding(0, 1),
 
+		InputModeReadonly: lipgloss.NewStyle().
+			Bold(true).
+			Foreground(lipgloss.Color(t.Muted)).
+			Background(lipgloss.Color(t.InputBg)).
+			Padding(0, 1),
+
+		InputModeYolo: lipgloss.NewStyle().
+			Bold(true).
+			Foreground(lipgloss.Color(t.Error)).
+			Background(lipgloss.Color(t.InputBg)).
+			Padding(0, 1),
+
+		InputModeCommand: lipgloss.NewStyle().
+			Bold(true).
+			Foreground(lipgloss.Color(t.User)).
+			Background(lipgloss.Color(t.InputBg)).
+			Padding(0, 1),
+
 		StatusReady: lipgloss.NewStyle().
 			Foreground(lipgloss.Color(t.Assistant)),
 
@@ -251,8 +274,10 @@ func bubbleMessageStyle(textColor, backgroundColor string) lipgloss.Style {
 func applyTextareaTheme(input *textarea.Model, t theme) {
 	focused := textarea.Style{
 		Base: lipgloss.NewStyle().
-			Foreground(lipgloss.Color(t.Text)),
-		CursorLine: lipgloss.NewStyle(),
+			Foreground(lipgloss.Color(t.Text)).
+			Background(lipgloss.Color(t.InputBg)),
+		CursorLine: lipgloss.NewStyle().
+			Background(lipgloss.Color(t.InputBg)),
 		Placeholder: lipgloss.NewStyle().
 			Foreground(lipgloss.Color(t.Muted)),
 		Prompt: lipgloss.NewStyle().
@@ -263,7 +288,7 @@ func applyTextareaTheme(input *textarea.Model, t theme) {
 			Foreground(lipgloss.Color(t.Surface)),
 	}
 	blurred := focused
-	blurred.CursorLine = lipgloss.NewStyle()
+	blurred.CursorLine = lipgloss.NewStyle().Background(lipgloss.Color(t.InputBg))
 
 	input.FocusedStyle = focused
 	input.BlurredStyle = blurred

main.go

@@ -1,228 +1,15 @@
 package main
 
 import (
-	"bufio"
-	"context"
-	"errors"
-	"flag"
 	"fmt"
-	"net/http"
 	"os"
-	"os/signal"
-	"path/filepath"
-	"strings"
 
-	"agentu/internal/agent"
-	"agentu/internal/session"
-	"agentu/internal/tools"
-	"agentu/internal/tui"
-	"agentu/pkg/config"
-	"agentu/pkg/llm"
+	"agentu/internal/app"
 )
 
-const defaultConfigContent = `# agentu configuration
-# Replace the placeholder values below with your actual provider settings.
-# See agentu.example.yaml for the full reference.
-providers:
-  default:
-    base_url: https://api.openai.com
-    api_key: YOUR_API_KEY_HERE
-    models:
-      - id: gpt-4o
-        context: 128000
-      - id: gpt-4o-mini
-        context: 128000
-`
-
 func main() {
-	if err := run(); err != nil {
+	if err := app.Run(); err != nil {
 		fmt.Fprintln(os.Stderr, "agentu:", err)
 		os.Exit(1)
 	}
 }
-
-func run() error {
-	configPath := flag.String("config", config.DefaultPath, "path to YAML config")
-	yolo := flag.Bool("yolo", false, "also enable automatic file writes and shell execution")
-	plain := flag.Bool("plain", false, "use the simple line-based REPL instead of the TUI")
-	themeName := flag.String("theme", "light", "TUI theme: light or dark")
-	resumeID := flag.String("resume", "", "resume a specific session by ID")
-	flag.Parse()
-
-	themeMode, err := tui.ParseThemeMode(*themeName)
-	if err != nil {
-		return err
-	}
-
-	cfg, err := config.Load(*configPath)
-	if err != nil {
-		if errors.Is(err, os.ErrNotExist) {
-			if bootstrapErr := bootstrapConfig(*configPath); bootstrapErr != nil {
-				return bootstrapErr
-			}
-			return fmt.Errorf("created default config at %s; edit it to set your API key and provider, then restart agentu", *configPath)
-		}
-		return err
-	}
-
-	providerName := cfg.DefaultProviderName()
-	providerConfig := cfg.Providers[providerName]
-	provider := llm.NewOpenAICompatibleClient(providerConfig.BaseURL, providerConfig.APIKey, http.DefaultClient)
-
-	var registry *tools.Registry
-	if *yolo {
-		registry = tools.Builtins(cfg.Agent.WorkingDir)
-	} else {
-		registry = tools.ReadOnlyBuiltins(cfg.Agent.WorkingDir)
-	}
-	registry.Register(tools.NewBuiltinSearchTool(http.DefaultClient))
-	registry.Register(tools.NewBuiltinFetchTool(http.DefaultClient))
-	systemPrompt := strings.TrimSpace(cfg.Agent.SystemPrompt + "\n\n" + tools.AgentInstructions(cfg.Agent.WorkingDir, *yolo))
-	maxContextTokens := cfg.Agent.MaxContextTokens
-	if maxContextTokens == 0 {
-		maxContextTokens = providerConfig.ContextTokens
-	}
-
-	assistant := agent.New(agent.Options{
-		Provider:         provider,
-		ProviderName:     providerName,
-		Model:            providerConfig.Model,
-		Thinking:         providerConfig.Thinking,
-		ThinkingParam:    providerConfig.ThinkingParam,
-		ThinkingEnabled:  providerConfig.ThinkingConfigured,
-		SystemPrompt:     systemPrompt,
-		ToolRegistry:     registry,
-		ToolTimeout:      cfg.Agent.ToolTimeout,
-		MaxContextTokens: maxContextTokens,
-	})
-
-	// Initialize session store
-	home, err := os.UserHomeDir()
-	if err != nil {
-		return fmt.Errorf("resolve home dir: %w", err)
-	}
-	sessionsDir := filepath.Join(home, ".agentu", "sessions")
-	store, err := session.NewStore(sessionsDir)
-	if err != nil {
-		return fmt.Errorf("init session store: %w", err)
-	}
-
-	modelManager, err := session.NewManager(cfg, assistant, http.DefaultClient, store)
-	if err != nil {
-		return err
-	}
-
-	if err := initializeSession(modelManager, *resumeID); err != nil {
-		return err
-	}
-
-	ctx, stop := signal.NotifyContext(context.Background(), os.Interrupt)
-	defer stop()
-
-	if !*plain {
-		err := tui.Run(ctx, assistant, tui.Options{
-			ModelName:    providerConfig.Model,
-			Yolo:         *yolo,
-			ThemeMode:    themeMode,
-			ModelManager: modelManager,
-			WorkingDir:   cfg.Agent.WorkingDir,
-		})
-		printExitSession(modelManager)
-		return err
-	}
-
-	return repl(ctx, assistant, modelManager)
-}
-
-func initializeSession(modelManager *session.Manager, resumeID string) error {
-	if resumeID != "" {
-		if err := modelManager.Resume(resumeID); err != nil {
-			return fmt.Errorf("resume session: %w", err)
-		}
-		return nil
-	}
-	if _, err := modelManager.NewSession(); err != nil {
-		return fmt.Errorf("start new session: %w", err)
-	}
-	return nil
-}
-
-func bootstrapConfig(configPath string) error {
-	resolved, err := config.ResolvePath(configPath)
-	if err != nil {
-		return fmt.Errorf("resolve config path: %w", err)
-	}
-	dir := filepath.Dir(resolved)
-	if err := os.MkdirAll(dir, 0o755); err != nil {
-		return fmt.Errorf("create config directory: %w", err)
-	}
-	if err := os.WriteFile(resolved, []byte(defaultConfigContent), 0o644); err != nil {
-		return fmt.Errorf("write default config: %w", err)
-	}
-	return nil
-}
-
-func repl(ctx context.Context, assistant *agent.Agent, modelManager *session.Manager) error {
-	scanner := bufio.NewScanner(os.Stdin)
-	scanner.Buffer(make([]byte, 0, 64*1024), 1024*1024)
-
-	fmt.Println("agentu")
-	fmt.Println("Type /exit to quit, /clear to reset context, /compact to compress context.")
-	for {
-		fmt.Print("> ")
-		if !scanner.Scan() {
-			if err := scanner.Err(); err != nil {
-				return err
-			}
-			fmt.Println()
-			printExitSession(modelManager)
-			return nil
-		}
-
-		input := strings.TrimSpace(scanner.Text())
-		switch input {
-		case "":
-			continue
-		case "/exit", "/quit":
-			printExitSession(modelManager)
-			return nil
-		case "/clear":
-			assistant.Clear()
-			fmt.Println("context cleared")
-			continue
-		case "/compact":
-			if err := assistant.Compact(ctx); err != nil {
-				fmt.Fprintln(os.Stderr, "compact error:", err)
-			} else {
-				fmt.Println("context compacted")
-				if modelManager != nil {
-					_ = modelManager.Save()
-				}
-			}
-			continue
-		}
-
-		if err := assistant.RunTurn(ctx, input, os.Stdout, os.Stderr); err != nil {
-			fmt.Fprintln(os.Stderr, "\nerror:", err)
-		}
-		_ = modelManager.Save()
-		fmt.Println()
-	}
-}
-
-func printExitSession(m *session.Manager) {
-	if m == nil {
-		return
-	}
-	_ = m.Save()
-	id := m.CurrentSessionID()
-	name := m.CurrentSessionName()
-	if id != "" {
-		if name != "" && name != id {
-			fmt.Fprintf(os.Stderr, "session: %s (%s)\n", id, name)
-		} else {
-			fmt.Fprintf(os.Stderr, "session: %s\n", id)
-		}
-		fmt.Fprintf(os.Stderr, "resume:  agentu --resume %s\n", id)
-	}
-}

pkg/config/config.go

@@ -14,6 +14,11 @@ import (
 
 const DefaultPath = "~/.agentu/config.yaml"
 
+const (
+	APITypeChat      = "chat"
+	APITypeResponses = "responses"
+)
+
 type Config struct {
 	Providers map[string]ProviderConfig `yaml:"providers"`
 	Agent     AgentConfig               `yaml:"agent"`
@@ -21,6 +26,7 @@ type Config struct {
 
 type ProviderConfig struct {
 	Name               string
+	APIType            string
 	BaseURL            string
 	Model              string
 	Models             []string
@@ -56,6 +62,7 @@ type rawConfig struct {
 }
 
 type rawProviderConfig struct {
+	APIType string           `yaml:"api_type"`
 	BaseURL string           `yaml:"base_url"`
 	Models  []rawModelConfig `yaml:"models"`
 	APIKey  string           `yaml:"api_key"`
@@ -153,6 +160,9 @@ func (c *Config) Validate() error {
 	for name, provider := range c.Providers {
 		var missing []string
 		prefix := "providers." + name
+		if !IsAPIType(provider.APIType) {
+			return fmt.Errorf("%s.api_type must be one of: %s", prefix, strings.Join(APITypes(), ", "))
+		}
 		if strings.TrimSpace(provider.BaseURL) == "" {
 			missing = append(missing, prefix+".base_url")
 		}
@@ -220,6 +230,7 @@ func normalizeProvider(name string, raw rawProviderConfig) ProviderConfig {
 	models, modelConfigs := normalizeModelConfigs(raw.Models)
 	provider := ProviderConfig{
 		Name:         name,
+		APIType:      normalizeAPIType(raw.APIType),
 		BaseURL:      raw.BaseURL,
 		Models:       models,
 		ModelConfigs: modelConfigs,
@@ -230,11 +241,44 @@ func normalizeProvider(name string, raw rawProviderConfig) ProviderConfig {
 	}
 	applyModelDefaults(&provider)
 	if provider.ThinkingConfigured && provider.ThinkingParam == "" {
-		provider.ThinkingParam = "thinking"
+		provider.ThinkingParam = DefaultThinkingParam(provider.APIType)
 	}
 	return provider
 }
 
+func normalizeAPIType(value string) string {
+	value = strings.ToLower(strings.TrimSpace(value))
+	if value == "" {
+		return APITypeChat
+	}
+	return value
+}
+
+func APITypes() []string {
+	return []string{APITypeChat, APITypeResponses}
+}
+
+func IsAPIType(value string) bool {
+	value = strings.ToLower(strings.TrimSpace(value))
+	for _, apiType := range APITypes() {
+		if value == apiType {
+			return true
+		}
+	}
+	return false
+}
+
+// DefaultThinkingParam returns the default request field used for the
+// thinking level. Chat-completions providers historically use a top-level
+// "thinking" field; the OpenAI Responses API uses "reasoning" with an
+// {"effort": ...} value.
+func DefaultThinkingParam(apiType string) string {
+	if apiType == APITypeResponses {
+		return "reasoning"
+	}
+	return "thinking"
+}
+
 func normalizeModelConfigs(rawModels []rawModelConfig) ([]string, map[string]ModelConfig) {
 	models := make([]string, 0, len(rawModels))
 	configs := make(map[string]ModelConfig, len(rawModels))

pkg/config/config_test.go

@@ -370,3 +370,110 @@ providers:
 		t.Fatalf("err = %v", err)
 	}
 }
+
+func TestLoadDefaultsAPITypeToChat(t *testing.T) {
+	dir := t.TempDir()
+	t.Setenv("AGENTU_TEST_KEY", "sk-test")
+	path := filepath.Join(dir, "agentu.yaml")
+	if err := os.WriteFile(path, []byte(`
+providers:
+  test:
+    base_url: https://example.com
+    api_key: ${AGENTU_TEST_KEY}
+    models:
+      - id: test-model
+`), 0o644); err != nil {
+		t.Fatal(err)
+	}
+
+	cfg, err := Load(path)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if got := cfg.Providers["test"].APIType; got != APITypeChat {
+		t.Fatalf("api type = %q, want %q", got, APITypeChat)
+	}
+}
+
+func TestLoadParsesResponsesAPIType(t *testing.T) {
+	dir := t.TempDir()
+	t.Setenv("AGENTU_TEST_KEY", "sk-test")
+	path := filepath.Join(dir, "agentu.yaml")
+	if err := os.WriteFile(path, []byte(`
+providers:
+  openai:
+    api_type: RESPONSES
+    base_url: https://api.openai.com
+    api_key: ${AGENTU_TEST_KEY}
+    models:
+      - id: gpt-test
+`), 0o644); err != nil {
+		t.Fatal(err)
+	}
+
+	cfg, err := Load(path)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if got := cfg.Providers["openai"].APIType; got != APITypeResponses {
+		t.Fatalf("api type = %q, want %q", got, APITypeResponses)
+	}
+}
+
+func TestLoadRejectsInvalidAPIType(t *testing.T) {
+	dir := t.TempDir()
+	t.Setenv("AGENTU_TEST_KEY", "sk-test")
+	path := filepath.Join(dir, "agentu.yaml")
+	if err := os.WriteFile(path, []byte(`
+providers:
+  test:
+    api_type: completions
+    base_url: https://example.com
+    api_key: ${AGENTU_TEST_KEY}
+    models:
+      - id: test-model
+`), 0o644); err != nil {
+		t.Fatal(err)
+	}
+
+	_, err := Load(path)
+	if err == nil || !strings.Contains(err.Error(), "api_type") {
+		t.Fatalf("err = %v", err)
+	}
+}
+
+func TestLoadDefaultsThinkingParamByAPIType(t *testing.T) {
+	dir := t.TempDir()
+	t.Setenv("AGENTU_TEST_KEY", "sk-test")
+	path := filepath.Join(dir, "agentu.yaml")
+	if err := os.WriteFile(path, []byte(`
+providers:
+  chat:
+    api_type: chat
+    base_url: https://chat.example.com
+    api_key: ${AGENTU_TEST_KEY}
+    models:
+      - id: chat-model
+        thinking: none
+  responses:
+    api_type: responses
+    base_url: https://responses.example.com
+    api_key: ${AGENTU_TEST_KEY}
+    models:
+      - id: responses-model
+        thinking: high
+`), 0o644); err != nil {
+		t.Fatal(err)
+	}
+
+	cfg, err := Load(path)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if got := cfg.Providers["chat"].ThinkingParam; got != "thinking" {
+		t.Fatalf("chat thinking param = %q, want %q", got, "thinking")
+	}
+	if got := cfg.Providers["responses"].ThinkingParam; got != "reasoning" {
+		t.Fatalf("responses thinking param = %q, want %q", got, "reasoning")
+	}
+}

pkg/llm/responses.go

@@ -0,0 +1,481 @@
+package llm
+
+import (
+	"bufio"
+	"bytes"
+	"context"
+	"encoding/json"
+	"errors"
+	"fmt"
+	"io"
+	"net/http"
+	"sort"
+	"strings"
+)
+
+const responsesEndpoint = "/v1/responses"
+
+// OpenAIResponsesClient implements llm.Provider against the OpenAI Responses
+// API (/v1/responses). It maps the agent's chat-style conversation history to
+// Responses input items (messages, function_call, function_call_output) and
+// converts Responses streaming events back into the shared StreamEvent shape.
+type OpenAIResponsesClient struct {
+	baseURL    string
+	apiKey     string
+	httpClient *http.Client
+}
+
+func NewOpenAIResponsesClient(baseURL, apiKey string, httpClient *http.Client) *OpenAIResponsesClient {
+	if httpClient == nil {
+		httpClient = http.DefaultClient
+	}
+	return &OpenAIResponsesClient{
+		baseURL:    strings.TrimRight(baseURL, "/"),
+		apiKey:     apiKey,
+		httpClient: httpClient,
+	}
+}
+
+func (c *OpenAIResponsesClient) ChatStream(ctx context.Context, req ChatRequest, emit func(StreamEvent) error) error {
+	body, err := json.Marshal(buildResponsesRequest(req))
+	if err != nil {
+		return fmt.Errorf("marshal responses request: %w", err)
+	}
+	state := &responsesStreamState{}
+	return postResponsesSSE(ctx, c.baseURL, c.apiKey, c.httpClient, body, func(payload string) error {
+		return state.handle(payload, emit)
+	})
+}
+
+func postResponsesSSE(ctx context.Context, baseURL, apiKey string, httpClient *http.Client, body []byte, handle func(payload string) error) error {
+	var lastErr error
+	for attempt := 1; attempt <= maxChatAttempts; attempt++ {
+		httpReq, err := http.NewRequestWithContext(ctx, http.MethodPost, baseURL+responsesEndpoint, bytes.NewReader(body))
+		if err != nil {
+			return fmt.Errorf("create responses request: %w", err)
+		}
+		httpReq.Header.Set("Content-Type", "application/json")
+		httpReq.Header.Set("Accept", "text/event-stream")
+		if apiKey != "" {
+			httpReq.Header.Set("Authorization", "Bearer "+apiKey)
+		}
+
+		resp, err := httpClient.Do(httpReq)
+		if err != nil {
+			lastErr = fmt.Errorf("send responses request: %w", err)
+			if !shouldRetryRequest(ctx, attempt, 0) {
+				return lastErr
+			}
+			if waitErr := waitBeforeRetry(ctx); waitErr != nil {
+				return waitErr
+			}
+			continue
+		}
+
+		if resp.StatusCode < 200 || resp.StatusCode >= 300 {
+			limit := io.LimitReader(resp.Body, 4096)
+			data, _ := io.ReadAll(limit)
+			_ = resp.Body.Close()
+			lastErr = fmt.Errorf("responses request failed: status=%d body=%s", resp.StatusCode, strings.TrimSpace(string(data)))
+			if !shouldRetryRequest(ctx, attempt, resp.StatusCode) {
+				return lastErr
+			}
+			if waitErr := waitBeforeRetry(ctx); waitErr != nil {
+				return waitErr
+			}
+			continue
+		}
+
+		defer resp.Body.Close()
+		return readResponsesSSE(resp.Body, handle)
+	}
+	return lastErr
+}
+
+func readResponsesSSE(r io.Reader, handle func(payload string) error) error {
+	reader := bufio.NewReader(r)
+	for {
+		line, err := reader.ReadString('\n')
+		if err != nil && !errors.Is(err, io.EOF) {
+			return fmt.Errorf("read responses stream: %w", err)
+		}
+
+		line = strings.TrimSpace(line)
+		if strings.HasPrefix(line, "data:") {
+			payload := strings.TrimSpace(strings.TrimPrefix(line, "data:"))
+			if payload == "[DONE]" {
+				return nil
+			}
+			if handleErr := handle(payload); handleErr != nil {
+				return handleErr
+			}
+		}
+
+		if errors.Is(err, io.EOF) {
+			return nil
+		}
+	}
+}
+
+// buildResponsesRequest converts a ChatRequest into the /v1/responses body.
+func buildResponsesRequest(req ChatRequest) map[string]any {
+	payload := map[string]any{
+		"model":  req.Model,
+		"input":  buildResponsesInput(req.Messages),
+		"stream": true,
+	}
+	if instructions := responsesInstructions(req.Messages); instructions != "" {
+		payload["instructions"] = instructions
+	}
+	if len(req.Tools) > 0 {
+		payload["tools"] = buildResponsesTools(req.Tools)
+	}
+	if req.ToolChoice != "" {
+		payload["tool_choice"] = req.ToolChoice
+	}
+	for key, value := range req.Extra {
+		if _, exists := payload[key]; exists {
+			continue
+		}
+		if key == "reasoning" {
+			if level, ok := value.(string); ok {
+				payload[key] = map[string]any{"effort": mapReasoningEffort(level)}
+				continue
+			}
+		}
+		payload[key] = value
+	}
+	return payload
+}
+
+// buildResponsesInput maps chat-style messages to Responses input items.
+// Assistant tool calls become function_call items and tool results become
+// function_call_output items, preserving their position in the conversation.
+func buildResponsesInput(messages []Message) []any {
+	items := make([]any, 0, len(messages)+4)
+	for _, msg := range messages {
+		switch msg.Role {
+		case RoleSystem:
+			// System instructions are sent as the top-level "instructions"
+			// field and must not be duplicated inside the input array.
+			continue
+		case RoleUser:
+			items = append(items, responsesMessageInput{Role: "user", Content: msg.Content})
+		case RoleAssistant:
+			if msg.Content != "" {
+				items = append(items, responsesMessageInput{Role: "assistant", Content: msg.Content})
+			}
+			for _, call := range msg.ToolCalls {
+				items = append(items, responsesFunctionCallInput{
+					Type:      "function_call",
+					CallID:    call.ID,
+					Name:      call.Function.Name,
+					Arguments: call.Function.Arguments,
+				})
+			}
+		case RoleTool:
+			items = append(items, responsesFunctionCallOutput{
+				Type:   "function_call_output",
+				CallID: msg.ToolCallID,
+				Output: msg.Content,
+			})
+		}
+	}
+	return items
+}
+
+// responsesInstructions extracts system messages into a single instructions
+// string, which is the OpenAI Responses API way to pass developer/system
+// guidance (and the most prompt-cache friendly).
+func responsesInstructions(messages []Message) string {
+	var instructions []string
+	for _, msg := range messages {
+		if msg.Role != RoleSystem {
+			continue
+		}
+		if text := strings.TrimSpace(msg.Content); text != "" {
+			instructions = append(instructions, text)
+		}
+	}
+	return strings.Join(instructions, "\n\n")
+}
+
+func buildResponsesTools(tools []Tool) []responsesTool {
+	out := make([]responsesTool, 0, len(tools))
+	for _, tool := range tools {
+		out = append(out, responsesTool{
+			Type:        "function",
+			Name:        tool.Function.Name,
+			Description: tool.Function.Description,
+			Parameters:  tool.Function.Parameters,
+		})
+	}
+	return out
+}
+
+func mapReasoningEffort(level string) string {
+	level = strings.ToLower(strings.TrimSpace(level))
+	if level == "middle" {
+		return "medium"
+	}
+	return level
+}
+
+type responsesMessageInput struct {
+	Role    string `json:"role"`
+	Content string `json:"content"`
+}
+
+type responsesFunctionCallInput struct {
+	Type      string `json:"type"`
+	CallID    string `json:"call_id"`
+	Name      string `json:"name"`
+	Arguments string `json:"arguments"`
+}
+
+type responsesFunctionCallOutput struct {
+	Type   string `json:"type"`
+	CallID string `json:"call_id"`
+	Output string `json:"output"`
+}
+
+type responsesTool struct {
+	Type        string          `json:"type"`
+	Name        string          `json:"name"`
+	Description string          `json:"description,omitempty"`
+	Parameters  json.RawMessage `json:"parameters,omitempty"`
+}
+
+// responsesStreamState tracks in-flight output items while parsing SSE events.
+type responsesStreamState struct {
+	calls map[int]*responsesCallBuilder
+	done  bool
+}
+
+type responsesCallBuilder struct {
+	index     int
+	id        string
+	callType  string
+	name      string
+	arguments strings.Builder
+	finalized bool
+}
+
+func (s *responsesStreamState) handle(payload string, emit func(StreamEvent) error) error {
+	if s.done {
+		return nil
+	}
+	var event responsesStreamEvent
+	if err := json.Unmarshal([]byte(payload), &event); err != nil {
+		return fmt.Errorf("parse responses stream payload: %w", err)
+	}
+
+	switch event.Type {
+	case "error":
+		s.done = true
+		return event.err()
+	case "response.failed":
+		s.done = true
+		if event.Response != nil && event.Response.Error != nil {
+			return fmt.Errorf("provider error: %s: %s", event.Response.Error.Code, event.Response.Error.Message)
+		}
+		return fmt.Errorf("provider error: response failed")
+	case "response.incomplete":
+		s.done = true
+		return s.finish(emit, "length", nil)
+	case "response.completed":
+		s.done = true
+		var usage *Usage
+		if event.Response != nil && event.Response.Usage != nil {
+			usage = &Usage{
+				PromptTokens:     event.Response.Usage.InputTokens,
+				CompletionTokens: event.Response.Usage.OutputTokens,
+				TotalTokens:      event.Response.Usage.TotalTokens,
+			}
+		}
+		return s.finish(emit, "stop", usage)
+	case "response.output_text.delta":
+		if event.Delta != "" {
+			return emit(StreamEvent{Content: event.Delta})
+		}
+	case "response.output_item.added":
+		if event.Item != nil && event.Item.Type == "function_call" {
+			s.builder(event.OutputIndex).applyAdded(*event.Item)
+		}
+	case "response.function_call_arguments.delta":
+		if event.Delta != "" {
+			s.builder(event.OutputIndex).arguments.WriteString(event.Delta)
+		}
+	case "response.function_call_arguments.done":
+		builder := s.builder(event.OutputIndex)
+		if event.Name != "" {
+			builder.name = event.Name
+		}
+		if event.Arguments != "" {
+			builder.arguments.Reset()
+			builder.arguments.WriteString(event.Arguments)
+		}
+	case "response.output_item.done":
+		if event.Item != nil && event.Item.Type == "function_call" {
+			builder := s.builder(event.OutputIndex)
+			builder.applyDone(*event.Item)
+			return s.emitCall(builder, emit)
+		}
+	}
+	return nil
+}
+
+// finish flushes any not-yet-finalized function calls and reports the
+// terminal finish reason.
+func (s *responsesStreamState) finish(emit func(StreamEvent) error, finishReason string, usage *Usage) error {
+	if len(s.calls) == 0 {
+		return emit(StreamEvent{FinishReason: finishReason, Usage: usage})
+	}
+	indexes := make([]int, 0, len(s.calls))
+	for index := range s.calls {
+		indexes = append(indexes, index)
+	}
+	sort.Ints(indexes)
+	var deltas []ToolCallDelta
+	for _, index := range indexes {
+		builder := s.calls[index]
+		if builder.finalized {
+			continue
+		}
+		builder.finalized = true
+		deltas = append(deltas, builder.delta())
+	}
+	if len(deltas) > 0 {
+		if err := emit(StreamEvent{ToolCalls: deltas}); err != nil {
+			return err
+		}
+	}
+	return emit(StreamEvent{FinishReason: finishReason, Usage: usage})
+}
+
+func (s *responsesStreamState) emitCall(builder *responsesCallBuilder, emit func(StreamEvent) error) error {
+	if builder.finalized {
+		return nil
+	}
+	builder.finalized = true
+	return emit(StreamEvent{ToolCalls: []ToolCallDelta{builder.delta()}})
+}
+
+func (s *responsesStreamState) builder(index int) *responsesCallBuilder {
+	if s.calls == nil {
+		s.calls = make(map[int]*responsesCallBuilder)
+	}
+	builder := s.calls[index]
+	if builder == nil {
+		builder = &responsesCallBuilder{index: index}
+		s.calls[index] = builder
+	}
+	return builder
+}
+
+func (b *responsesCallBuilder) applyAdded(item responsesStreamItem) {
+	if b.id == "" {
+		b.id = item.CallID
+		if b.id == "" {
+			b.id = item.ID
+		}
+	}
+	if b.name == "" {
+		b.name = item.Name
+	}
+	if item.Arguments != "" {
+		b.arguments.Reset()
+		b.arguments.WriteString(item.Arguments)
+	}
+}
+
+func (b *responsesCallBuilder) applyDone(item responsesStreamItem) {
+	if item.CallID != "" {
+		b.id = item.CallID
+	} else if item.ID != "" {
+		b.id = item.ID
+	}
+	if item.Name != "" {
+		b.name = item.Name
+	}
+	if item.Arguments != "" {
+		b.arguments.Reset()
+		b.arguments.WriteString(item.Arguments)
+	}
+}
+
+func (b *responsesCallBuilder) delta() ToolCallDelta {
+	id := b.id
+	if id == "" {
+		id = fmt.Sprintf("call_%d", b.index)
+	}
+	callType := b.callType
+	if callType == "" {
+		callType = "function"
+	}
+	return ToolCallDelta{
+		Index:     b.index,
+		ID:        id,
+		Type:      callType,
+		Name:      b.name,
+		Arguments: b.arguments.String(),
+	}
+}
+
+type responsesStreamEvent struct {
+	Type        string                   `json:"type"`
+	OutputIndex int                      `json:"output_index"`
+	Delta       string                   `json:"delta"`
+	Arguments   string                   `json:"arguments"`
+	Name        string                   `json:"name"`
+	Code        string                   `json:"code"`
+	Message     string                   `json:"message"`
+	Param       string                   `json:"param"`
+	Item        *responsesStreamItem     `json:"item"`
+	Response    *responsesStreamResponse `json:"response"`
+	Error       *responsesStreamError    `json:"error"`
+}
+
+type responsesStreamItem struct {
+	Type      string `json:"type"`
+	ID        string `json:"id"`
+	CallID    string `json:"call_id"`
+	Name      string `json:"name"`
+	Arguments string `json:"arguments"`
+}
+
+type responsesStreamResponse struct {
+	Status string                `json:"status"`
+	Error  *responsesStreamError `json:"error"`
+	Usage  *responsesStreamUsage `json:"usage"`
+}
+
+type responsesStreamUsage struct {
+	InputTokens  int `json:"input_tokens"`
+	OutputTokens int `json:"output_tokens"`
+	TotalTokens  int `json:"total_tokens"`
+}
+
+type responsesStreamError struct {
+	Type    string `json:"type"`
+	Code    string `json:"code"`
+	Message string `json:"message"`
+	Param   string `json:"param"`
+}
+
+func (e responsesStreamEvent) err() error {
+	code, message, param := e.Code, e.Message, e.Param
+	if code == "" && message == "" && e.Error != nil {
+		code, message, param = e.Error.Code, e.Error.Message, e.Error.Param
+	}
+	if message == "" {
+		if param != "" {
+			return fmt.Errorf("provider error: %s", param)
+		}
+		return fmt.Errorf("provider error")
+	}
+	if code != "" {
+		return fmt.Errorf("provider error: %s: %s", code, message)
+	}
+	return fmt.Errorf("provider error: %s", message)
+}

pkg/llm/responses_test.go

@@ -0,0 +1,337 @@
+package llm
+
+import (
+	"context"
+	"encoding/json"
+	"net/http"
+	"net/http/httptest"
+	"strings"
+	"testing"
+)
+
+func TestBuildResponsesRequest(t *testing.T) {
+	req := ChatRequest{
+		Model: "gpt-test",
+		Messages: []Message{
+			{Role: RoleSystem, Content: "be concise"},
+			{Role: RoleUser, Content: "hi"},
+			{
+				Role:    RoleAssistant,
+				Content: "let me check",
+				ToolCalls: []ToolCall{{
+					ID:   "call_1",
+					Type: "function",
+					Function: FunctionCall{
+						Name:      "file_read",
+						Arguments: `{"path":"README.md"}`,
+					},
+				}},
+			},
+			{Role: RoleTool, ToolCallID: "call_1", Content: "contents"},
+			{Role: RoleAssistant, Content: "done"},
+		},
+		Tools: []Tool{{
+			Type: "function",
+			Function: ToolFunction{
+				Name:        "file_read",
+				Description: "Read a local file",
+				Parameters:  json.RawMessage(`{"type":"object","properties":{}}`),
+			},
+		}},
+		ToolChoice: "auto",
+		Extra: map[string]any{
+			"reasoning":   "middle",
+			"thinking":    "high",
+			"temperature": 0.2,
+		},
+	}
+
+	payload := buildResponsesRequest(req)
+	data, err := json.Marshal(payload)
+	if err != nil {
+		t.Fatal(err)
+	}
+	var raw map[string]any
+	if err := json.Unmarshal(data, &raw); err != nil {
+		t.Fatal(err)
+	}
+
+	if raw["model"] != "gpt-test" || raw["stream"] != true {
+		t.Fatalf("model/stream = %#v / %#v", raw["model"], raw["stream"])
+	}
+	if raw["instructions"] != "be concise" {
+		t.Fatalf("instructions = %#v", raw["instructions"])
+	}
+	if raw["tool_choice"] != "auto" {
+		t.Fatalf("tool_choice = %#v", raw["tool_choice"])
+	}
+
+	input, ok := raw["input"].([]any)
+	if !ok || len(input) != 5 {
+		t.Fatalf("input = %#v", raw["input"])
+	}
+	checkInputItem := func(index int, wantType, wantKey, wantValue string) {
+		t.Helper()
+		item, ok := input[index].(map[string]any)
+		if !ok {
+			t.Fatalf("input[%d] = %#v", index, input[index])
+		}
+		if item[wantKey] != wantValue {
+			t.Fatalf("input[%d].%s = %#v, want %q", index, wantKey, item[wantKey], wantValue)
+		}
+		if wantType != "" && item["type"] != wantType {
+			t.Fatalf("input[%d].type = %#v, want %q", index, item["type"], wantType)
+		}
+	}
+	checkInputItem(0, "", "role", "user")
+	checkInputItem(1, "", "role", "assistant")
+	checkInputItem(2, "function_call", "call_id", "call_1")
+	checkInputItem(3, "function_call_output", "call_id", "call_1")
+	checkInputItem(4, "", "role", "assistant")
+	if item := input[2].(map[string]any); item["name"] != "file_read" || item["arguments"] != `{"path":"README.md"}` {
+		t.Fatalf("function_call item = %#v", item)
+	}
+	if item := input[3].(map[string]any); item["output"] != "contents" {
+		t.Fatalf("function_call_output item = %#v", item)
+	}
+
+	tools, ok := raw["tools"].([]any)
+	if !ok || len(tools) != 1 {
+		t.Fatalf("tools = %#v", raw["tools"])
+	}
+	tool := tools[0].(map[string]any)
+	if tool["type"] != "function" || tool["name"] != "file_read" || tool["description"] != "Read a local file" {
+		t.Fatalf("tool = %#v", tool)
+	}
+	if _, ok := tool["parameters"].(map[string]any); !ok {
+		t.Fatalf("tool parameters = %#v", tool["parameters"])
+	}
+
+	reasoning, ok := raw["reasoning"].(map[string]any)
+	if !ok || reasoning["effort"] != "medium" {
+		t.Fatalf("reasoning = %#v", raw["reasoning"])
+	}
+	if raw["thinking"] != "high" {
+		t.Fatalf("thinking = %#v", raw["thinking"])
+	}
+	if raw["temperature"] != 0.2 {
+		t.Fatalf("temperature = %#v", raw["temperature"])
+	}
+}
+
+func TestResponsesInstructionsCombinesSystemMessages(t *testing.T) {
+	payload := buildResponsesRequest(ChatRequest{
+		Model: "gpt-test",
+		Messages: []Message{
+			{Role: RoleSystem, Content: "first instruction"},
+			{Role: RoleUser, Content: "hi"},
+			{Role: RoleSystem, Content: "second instruction"},
+		},
+	})
+	data, err := json.Marshal(payload)
+	if err != nil {
+		t.Fatal(err)
+	}
+	var raw map[string]any
+	if err := json.Unmarshal(data, &raw); err != nil {
+		t.Fatal(err)
+	}
+	if raw["instructions"] != "first instruction\n\nsecond instruction" {
+		t.Fatalf("instructions = %#v", raw["instructions"])
+	}
+	input, ok := raw["input"].([]any)
+	if !ok || len(input) != 1 {
+		t.Fatalf("input = %#v", raw["input"])
+	}
+	item, ok := input[0].(map[string]any)
+	if !ok || item["role"] != "user" {
+		t.Fatalf("input[0] = %#v", input[0])
+	}
+}
+
+func TestOpenAIResponsesClientStreamsContentAndToolCalls(t *testing.T) {
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		if r.URL.Path != "/v1/responses" {
+			t.Fatalf("path = %s", r.URL.Path)
+		}
+		if got := r.Header.Get("Authorization"); got != "Bearer sk-test" {
+			t.Fatalf("authorization = %q", got)
+		}
+		body := mustReadBody(t, r)
+		var raw map[string]any
+		if err := json.NewDecoder(strings.NewReader(body)).Decode(&raw); err != nil {
+			t.Fatal(err)
+		}
+		if raw["stream"] != true {
+			t.Fatal("request did not enable stream")
+		}
+		if raw["tool_choice"] != "auto" {
+			t.Fatalf("tool_choice = %#v", raw["tool_choice"])
+		}
+		input, ok := raw["input"].([]any)
+		if !ok || len(input) == 0 {
+			t.Fatalf("input = %#v", raw["input"])
+		}
+
+		w.Header().Set("Content-Type", "text/event-stream")
+		events := []string{
+			`{"type":"response.output_item.added","output_index":0,"item":{"type":"message","id":"msg_1","role":"assistant","content":[]}}`,
+			`{"type":"response.output_text.delta","item_id":"msg_1","output_index":0,"content_index":0,"delta":"Hel"}`,
+			`{"type":"response.output_text.delta","item_id":"msg_1","output_index":0,"content_index":0,"delta":"lo"}`,
+			`{"type":"response.output_item.added","output_index":1,"item":{"type":"function_call","id":"fc_1","call_id":"call_1","name":"file_read","arguments":""}}`,
+			`{"type":"response.function_call_arguments.delta","item_id":"fc_1","output_index":1,"delta":"{\"path\""}`,
+			`{"type":"response.function_call_arguments.delta","item_id":"fc_1","output_index":1,"delta":":\"README.md\"}"}`,
+			`{"type":"response.function_call_arguments.done","item_id":"fc_1","output_index":1,"name":"file_read","arguments":"{\"path\":\"README.md\"}"}`,
+			`{"type":"response.output_item.done","output_index":1,"item":{"type":"function_call","id":"fc_1","call_id":"call_1","name":"file_read","arguments":"{\"path\":\"README.md\"}"}}`,
+			`{"type":"response.output_item.done","output_index":0,"item":{"type":"message","id":"msg_1","role":"assistant","content":[{"type":"output_text","text":"Hello"}]}}`,
+			`{"type":"response.completed","response":{"id":"resp_1","status":"completed","usage":{"input_tokens":10,"output_tokens":20,"total_tokens":30}}}`,
+		}
+		for _, event := range events {
+			_, _ = w.Write([]byte("data: " + event + "\n\n"))
+		}
+	}))
+	defer server.Close()
+
+	client := NewOpenAIResponsesClient(server.URL, "sk-test", server.Client())
+	var content strings.Builder
+	var calls []ToolCallDelta
+	var finishes []string
+	var usage *Usage
+	err := client.ChatStream(context.Background(), ChatRequest{
+		Model:      "gpt-test",
+		Messages:   []Message{{Role: RoleUser, Content: "hi"}},
+		ToolChoice: "auto",
+		Tools: []Tool{{
+			Type: "function",
+			Function: ToolFunction{
+				Name:        "file_read",
+				Description: "Read a local file",
+				Parameters:  json.RawMessage(`{"type":"object"}`),
+			},
+		}},
+	}, func(event StreamEvent) error {
+		content.WriteString(event.Content)
+		calls = append(calls, event.ToolCalls...)
+		if event.Usage != nil {
+			usage = event.Usage
+		}
+		if event.FinishReason != "" {
+			finishes = append(finishes, event.FinishReason)
+		}
+		return nil
+	})
+	if err != nil {
+		t.Fatal(err)
+	}
+	if content.String() != "Hello" {
+		t.Fatalf("content = %q", content.String())
+	}
+	if len(calls) != 1 {
+		t.Fatalf("calls = %#v", calls)
+	}
+	call := calls[0]
+	if call.Index != 1 || call.ID != "call_1" || call.Name != "file_read" {
+		t.Fatalf("call = %#v", call)
+	}
+	if call.Arguments != `{"path":"README.md"}` {
+		t.Fatalf("arguments = %q", call.Arguments)
+	}
+	if len(finishes) != 1 || finishes[0] != "stop" {
+		t.Fatalf("finishes = %#v", finishes)
+	}
+	if usage == nil || usage.PromptTokens != 10 || usage.CompletionTokens != 20 || usage.TotalTokens != 30 {
+		t.Fatalf("usage = %#v", usage)
+	}
+}
+
+func TestOpenAIResponsesClientErrorEvent(t *testing.T) {
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		w.Header().Set("Content-Type", "text/event-stream")
+		_, _ = w.Write([]byte(`data: {"type":"error","code":"invalid_request_error","message":"bad request"}` + "\n\n"))
+	}))
+	defer server.Close()
+
+	client := NewOpenAIResponsesClient(server.URL, "sk-test", server.Client())
+	err := client.ChatStream(context.Background(), ChatRequest{Model: "test"}, func(StreamEvent) error {
+		return nil
+	})
+	if err == nil || !strings.Contains(err.Error(), "bad request") {
+		t.Fatalf("err = %v", err)
+	}
+}
+
+func TestOpenAIResponsesClientFailedEvent(t *testing.T) {
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		w.Header().Set("Content-Type", "text/event-stream")
+		_, _ = w.Write([]byte(`data: {"type":"response.failed","response":{"id":"resp_1","status":"failed","error":{"code":"server_error","message":"boom"}}}` + "\n\n"))
+	}))
+	defer server.Close()
+
+	client := NewOpenAIResponsesClient(server.URL, "sk-test", server.Client())
+	err := client.ChatStream(context.Background(), ChatRequest{Model: "test"}, func(StreamEvent) error {
+		return nil
+	})
+	if err == nil || !strings.Contains(err.Error(), "boom") {
+		t.Fatalf("err = %v", err)
+	}
+}
+
+func TestOpenAIResponsesClientIncomplete(t *testing.T) {
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		w.Header().Set("Content-Type", "text/event-stream")
+		_, _ = w.Write([]byte(`data: {"type":"response.output_text.delta","item_id":"msg_1","output_index":0,"content_index":0,"delta":"partial"}` + "\n\n"))
+		_, _ = w.Write([]byte(`data: {"type":"response.incomplete","response":{"id":"resp_1","status":"incomplete","incomplete_details":{"reason":"max_tokens"}}}` + "\n\n"))
+	}))
+	defer server.Close()
+
+	client := NewOpenAIResponsesClient(server.URL, "sk-test", server.Client())
+	var content strings.Builder
+	var finishes []string
+	err := client.ChatStream(context.Background(), ChatRequest{Model: "test"}, func(event StreamEvent) error {
+		content.WriteString(event.Content)
+		if event.FinishReason != "" {
+			finishes = append(finishes, event.FinishReason)
+		}
+		return nil
+	})
+	if err != nil {
+		t.Fatal(err)
+	}
+	if content.String() != "partial" {
+		t.Fatalf("content = %q", content.String())
+	}
+	if len(finishes) != 1 || finishes[0] != "length" {
+		t.Fatalf("finishes = %#v", finishes)
+	}
+}
+
+func TestOpenAIResponsesClientRetriesTransientHTTPError(t *testing.T) {
+	attempts := 0
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		attempts++
+		if attempts == 1 {
+			http.Error(w, "temporary", http.StatusBadGateway)
+			return
+		}
+		w.Header().Set("Content-Type", "text/event-stream")
+		_, _ = w.Write([]byte(`data: {"type":"response.output_text.delta","item_id":"msg_1","output_index":0,"content_index":0,"delta":"ok"}` + "\n\n"))
+		_, _ = w.Write([]byte(`data: {"type":"response.completed","response":{"id":"resp_1","status":"completed"}}` + "\n\n"))
+	}))
+	defer server.Close()
+
+	client := NewOpenAIResponsesClient(server.URL, "sk-test", server.Client())
+	var content strings.Builder
+	err := client.ChatStream(context.Background(), ChatRequest{Model: "test"}, func(event StreamEvent) error {
+		content.WriteString(event.Content)
+		return nil
+	})
+	if err != nil {
+		t.Fatal(err)
+	}
+	if attempts != 2 {
+		t.Fatalf("attempts = %d, want 2", attempts)
+	}
+	if content.String() != "ok" {
+		t.Fatalf("content = %q", content.String())
+	}
+}