[测试、性能、安全、验收与归档]：完成全量测试验收与 OpenSpec 归档，包含 CLI 验收、NewAPI smoke、安全扫描和主 spec 更新

2026-06-17 18:42:11 +08:00
parent 667c8941f2
commit 1b884ac6a9
15 changed files with 522 additions and 28 deletions
@@ -1,6 +1,6 @@
 OWNER_LLM_BASE_URL=https://newapi.mkbk.shop
 OWNER_LLM_API_KEY=
-OWNER_LLM_MODEL=gpt-4o-mini
+OWNER_LLM_MODEL=gpt-5.4-mini
 OWNER_LLM_API_STYLE=chat_completions
 OWNER_AUDIO_INPUT_DEVICE=
 OWNER_AUDIO_OUTPUT_DEVICE=
@@ -8,6 +8,7 @@ __pycache__/
 .pytest_cache/
 .mypy_cache/
 .ruff_cache/
+.serena/
 build/
 dist/
 *.egg-info/
@@ -43,9 +43,9 @@

 ## 6. 测试、性能、安全、验收与归档

- [ ] 6.1 完成自动化测试套件；前置条件：所有模块已实现；验收标准：覆盖核心模型、配置、Transport、Wake/VAD/STT、Context、LLM/TTS、Pipeline、UI、资产；测试要点：`python3.11 -m unittest discover -s tests` 通过；优先级：P0；预计：60 分钟。
- [ ] 6.2 完成端到端验收命令；前置条件：Pipeline 可用；验收标准：fixture 音频链路可从“小杰小杰”走到播放输出和上下文更新；测试要点：CLI acceptance 命令退出码为 0；优先级：P0；预计：45 分钟。
- [ ] 6.3 完成 OpenAI/NewAPI smoke 验收；前置条件：提供临时 API key 环境变量；验收标准：调用配置的 base URL 获得非空 LLM 回复，失败时输出结构化诊断且不泄露 key；测试要点：真实 smoke 或可解释的网络失败证据；优先级：P0；预计：45 分钟。
- [ ] 6.4 完成安全检查；前置条件：全部代码已实现；验收标准：仓库中不包含 API key，日志脱敏，默认不持久化原始音频；测试要点：secret grep、配置测试和日志脱敏测试通过；优先级：P0；预计：30 分钟。
- [ ] 6.5 完成 OpenSpec archive；前置条件：所有任务完成并验证通过；验收标准：`openspec archive add-voice-pet-pipeline` 后主 spec 更新，change 进入 archive；测试要点：`openspec validate --all --strict` 通过；优先级：P0；预计：30 分钟。
- [ ] 6.6 完成最终验收提交；前置条件：6.1 至 6.5 已完成；验收标准：完整测试、OpenSpec strict 校验、git status 审计通过后立即 commit；测试要点：提交信息使用“`[测试、性能、安全、验收与归档]：完成[具体功能描述]，包含[关键变更]`”格式；优先级：P0；预计：20 分钟。
+- [x] 6.1 完成自动化测试套件；前置条件：所有模块已实现；验收标准：覆盖核心模型、配置、Transport、Wake/VAD/STT、Context、LLM/TTS、Pipeline、UI、资产；测试要点：`python3.11 -m unittest discover -s tests` 通过；优先级：P0；预计：60 分钟。
+- [x] 6.2 完成端到端验收命令；前置条件：Pipeline 可用；验收标准：fixture 音频链路可从“小杰小杰”走到播放输出和上下文更新；测试要点：CLI acceptance 命令退出码为 0；优先级：P0；预计：45 分钟。
+- [x] 6.3 完成 OpenAI/NewAPI smoke 验收；前置条件：提供临时 API key 环境变量；验收标准：调用配置的 base URL 获得非空 LLM 回复，失败时输出结构化诊断且不泄露 key；测试要点：真实 smoke 或可解释的网络失败证据；优先级：P0；预计：45 分钟。
+- [x] 6.4 完成安全检查；前置条件：全部代码已实现；验收标准：仓库中不包含 API key，日志脱敏，默认不持久化原始音频；测试要点：secret grep、配置测试和日志脱敏测试通过；优先级：P0；预计：30 分钟。
+- [x] 6.5 完成 OpenSpec archive；前置条件：所有任务完成并验证通过；验收标准：`openspec archive add-voice-pet-pipeline` 后主 spec 更新，change 进入 archive；测试要点：`openspec validate --all --strict` 通过；优先级：P0；预计：30 分钟。
+- [x] 6.6 完成最终验收提交；前置条件：6.1 至 6.5 已完成；验收标准：完整测试、OpenSpec strict 校验、git status 审计通过后立即 commit；测试要点：提交信息使用“`[测试、性能、安全、验收与归档]：完成[具体功能描述]，包含[关键变更]`”格式；优先级：P0；预计：20 分钟。
@@ -0,0 +1,268 @@
+# voice-pet-pipeline Specification
+
+## Purpose
+TBD - created by archiving change add-voice-pet-pipeline. Update Purpose after archive.
+## Requirements
+### Requirement: Python desktop pet runtime
+The system SHALL be specified as a local Python desktop pet application that owns the voice pipeline, desktop pet state, local audio input, and local audio output in the first version.
+
+#### Scenario: App starts in local desktop mode
+- **WHEN** the user starts the future application on macOS
+- **THEN** the system SHALL initialize as a local desktop pet process rather than a remote service or browser-only tool
+
+#### Scenario: Implementation follows OpenSpec artifacts
+- **WHEN** this OpenSpec change is implemented
+- **THEN** the repository SHALL contain Python source code, automated tests, validation commands, and project-local pet assets aligned with the planning artifacts
+
+### Requirement: Local microphone and speaker transport
+The system SHALL use the local microphone as the first-version input Transport and the local system speaker as the first-version output Transport.
+
+#### Scenario: Microphone input is available
+- **WHEN** the configured microphone is available and permitted
+- **THEN** the Transport SHALL provide streaming audio frames to the wakeword and VAD stages
+
+#### Scenario: Speaker output is available
+- **WHEN** TTS returns a valid audio segment and the configured speaker is available
+- **THEN** the Transport SHALL play the audio segment through the local speaker
+
+#### Scenario: Audio device is unavailable
+- **WHEN** the microphone or speaker is missing, denied, or unsupported
+- **THEN** the system SHALL expose a recoverable Transport error with a stable error code
+
+### Requirement: Wake word detection
+The system SHALL listen locally for the Chinese wake word “小杰小杰” before accepting user speech for a conversation turn.
+
+#### Scenario: Wake word is detected
+- **WHEN** the user says “小杰小杰” and the wakeword provider returns confidence above the configured threshold
+- **THEN** the pipeline SHALL transition from wake listening to speech detection
+
+#### Scenario: Wake word is not detected
+- **WHEN** background speech or noise does not match “小杰小杰”
+- **THEN** the pipeline SHALL remain in wake listening and SHALL NOT invoke STT, LLM, or TTS
+
+#### Scenario: Wake model fails
+- **WHEN** the wakeword provider cannot load or process audio
+- **THEN** the system SHALL report a wakeword error and SHALL NOT crash the desktop pet process
+
+### Requirement: VAD speech endpoint detection
+After wakeword detection, the system SHALL use VAD to identify when the user starts and stops speaking.
+
+#### Scenario: User begins speaking
+- **WHEN** VAD detects continuous speech above the configured start threshold
+- **THEN** the pipeline SHALL begin recording the current utterance
+
+#### Scenario: User stops speaking
+- **WHEN** VAD detects continuous silence above the configured end threshold
+- **THEN** the pipeline SHALL close the current audio segment and send it to STT
+
+#### Scenario: User says nothing after wakeword
+- **WHEN** no speech is detected before the configured no-speech timeout
+- **THEN** the pipeline SHALL return to wake listening without invoking STT or LLM
+
+#### Scenario: Recording exceeds maximum duration
+- **WHEN** speech continues beyond the configured maximum utterance duration
+- **THEN** the pipeline SHALL end the segment, mark the end reason, and continue to STT with the captured audio
+
+### Requirement: Local STT transcription
+The system SHALL transcribe the captured user utterance through a local STT provider, with `sherpa-onnx` as the recommended first implementation candidate.
+
+#### Scenario: STT succeeds
+- **WHEN** STT returns non-empty text for a captured audio segment
+- **THEN** the pipeline SHALL add the trimmed text as a user message to the conversation context
+
+#### Scenario: STT returns empty text
+- **WHEN** STT returns empty text, punctuation-only text, or an invalid transcript
+- **THEN** the pipeline SHALL skip LLM invocation and return to wake listening with a recoverable status
+
+#### Scenario: STT provider fails
+- **WHEN** the STT provider raises an error or cannot load its local model
+- **THEN** the system SHALL emit an STT error code and SHALL recover to a state where future wake attempts are possible
+
+### Requirement: Conversation context management
+The system SHALL maintain an in-memory conversation context for the active desktop pet session.
+
+#### Scenario: User transcript is accepted
+- **WHEN** a valid STT transcript is produced
+- **THEN** the system SHALL append it to the context as a user message before invoking the LLM
+
+#### Scenario: Assistant reply completes
+- **WHEN** the LLM and TTS stages complete a reply
+- **THEN** the system SHALL append the assistant text to the context
+
+#### Scenario: Context exceeds configured budget
+- **WHEN** the context exceeds the configured message, character, or token budget
+- **THEN** the system SHALL preserve the system prompt and most recent conversation turns while removing older ordinary messages
+
+### Requirement: Cloud LLM streaming reply
+The system SHALL use a cloud LLM provider for reply generation, with OpenAI Responses API streaming as the recommended first implementation.
+
+#### Scenario: LLM stream begins
+- **WHEN** the LLM provider receives valid context messages and configuration
+- **THEN** it SHALL stream reply deltas rather than waiting for the full reply before producing output
+
+#### Scenario: LLM model is configured
+- **WHEN** the system starts
+- **THEN** the LLM model name SHALL be read from configuration and SHALL NOT be hard-coded in pipeline logic
+
+#### Scenario: LLM request fails
+- **WHEN** the LLM provider times out, is rate limited, lacks credentials, or encounters a network error
+- **THEN** the pipeline SHALL emit a structured LLM error and transition through error recovery to wake listening
+
+### Requirement: Local TTS synthesis and playback
+The system SHALL synthesize assistant replies through a local TTS provider, with `sherpa-onnx` as the recommended first implementation candidate.
+
+#### Scenario: Sentence is ready for speech
+- **WHEN** the LLM stream produces a complete sentence or configured text chunk
+- **THEN** the TTS provider SHALL synthesize that text into an audio segment
+
+#### Scenario: First TTS segment is ready
+- **WHEN** the first synthesized audio segment is available
+- **THEN** the output Transport SHALL begin playback without waiting for every remaining LLM delta
+
+#### Scenario: TTS fails
+- **WHEN** TTS returns empty audio or raises an error
+- **THEN** the pipeline SHALL emit a structured TTS error and SHALL recover without terminating the application
+
+### Requirement: Pipeline state machine
+The system SHALL expose deterministic pipeline states for `idle`, `wake_listening`, `speech_detecting`, `recording`, `transcribing`, `thinking`, `speaking`, `interrupted`, and `error_recovering`.
+
+#### Scenario: Normal conversation turn
+- **WHEN** wakeword, VAD, STT, LLM, TTS, and playback all succeed
+- **THEN** the state sequence SHALL progress through wake listening, speech detection, recording, transcribing, thinking, speaking, and back to wake listening
+
+#### Scenario: Recoverable provider error
+- **WHEN** any provider fails during a conversation turn
+- **THEN** the state machine SHALL enter error recovery and then return to wake listening after cleanup
+
+#### Scenario: UI subscribes to state
+- **WHEN** the pipeline state changes
+- **THEN** the desktop pet UI SHALL be able to update its visible state without directly invoking audio or model providers
+
+### Requirement: Desktop pet visual states
+The system SHALL define desktop pet visual states for idle, listening, recording, thinking, speaking, and error feedback.
+
+#### Scenario: Wakeword is detected
+- **WHEN** the pipeline enters speech detection or recording
+- **THEN** the desktop pet SHALL show a listening or recording visual state
+
+#### Scenario: LLM is generating
+- **WHEN** the pipeline enters thinking
+- **THEN** the desktop pet SHALL show a thinking visual state
+
+#### Scenario: TTS is playing
+- **WHEN** the pipeline enters speaking
+- **THEN** the desktop pet SHALL show a speaking visual state
+
+#### Scenario: Error occurs
+- **WHEN** the pipeline enters error recovery
+- **THEN** the desktop pet SHALL show a concise visible error state and then return to idle or wake listening when recovered
+
+### Requirement: Generated pet asset specification
+The system SHALL specify that production desktop pet images are generated with the image generation skill when usable, or with a reproducible project-local fallback when image generation output is unavailable or fails validation, and saved inside the project as transparent PNG assets.
+
+#### Scenario: Image generation output is usable
+- **WHEN** pet images are generated with `imagegen` and pass role and transparency validation
+- **THEN** the images SHALL be processed for transparency, validated, and saved to a project asset directory
+
+#### Scenario: Image generation output is unavailable or invalid
+- **WHEN** image generation output is unavailable, inaccessible as a project file, or fails role/transparency validation
+- **THEN** the system SHALL use a reproducible project-local fallback to create transparent PNG pet state assets and SHALL validate them before use
+
+#### Scenario: Planning phase is executed
+- **WHEN** this OpenSpec change is implemented
+- **THEN** generated or generated-derived pet assets SHALL be saved inside the project and SHALL NOT be referenced from a temporary generation directory
+
+### Requirement: Audio feedback suppression
+The system SHALL prevent the desktop pet's own TTS playback from being treated as a new user wakeword or speech input.
+
+#### Scenario: TTS playback is active
+- **WHEN** the output Transport is playing synthesized assistant speech
+- **THEN** wakeword detection and VAD input processing SHALL be suppressed or ignored for that playback window by default
+
+#### Scenario: User speaks during playback
+- **WHEN** the user speaks while TTS playback is active in the first version
+- **THEN** the system SHALL ignore that speech unless a future interruption feature is explicitly specified
+
+### Requirement: Structured errors and logging
+The system SHALL represent stage failures with structured error codes and log state transitions, latency metrics, and provider names.
+
+#### Scenario: Stage transition occurs
+- **WHEN** the pipeline changes state
+- **THEN** the system SHALL log the turn ID, old state, new state, timestamp, and triggering event
+
+#### Scenario: Provider call completes
+- **WHEN** a provider call succeeds or fails
+- **THEN** the system SHALL log provider name, stage, duration, and error code if present
+
+#### Scenario: Sensitive data is present
+- **WHEN** logs are written
+- **THEN** the system SHALL NOT log API keys, authorization headers, or raw credentials
+
+### Requirement: Performance targets
+The system SHALL define measurable first-version latency and reliability targets for wakeword response, endpoint detection, LLM first output, and speech playback.
+
+#### Scenario: Wakeword latency is measured
+- **WHEN** wakeword detection succeeds in a normal local environment
+- **THEN** the system SHALL target visible listening feedback within 800 ms
+
+#### Scenario: User stops speaking
+- **WHEN** the user stops speaking after a normal utterance
+- **THEN** VAD endpoint detection SHALL target transition to transcription within 900 ms
+
+#### Scenario: LLM begins responding
+- **WHEN** a valid transcript has been sent to the LLM provider
+- **THEN** the system SHALL target first playable reply text within 3.5 seconds
+
+#### Scenario: Normal reply is spoken
+- **WHEN** a normal Chinese question under 10 seconds is processed
+- **THEN** the system SHALL target the start of spoken playback within 5 seconds P95 after the user stops speaking
+
+### Requirement: Security and privacy
+The system SHALL protect credentials and minimize audio persistence.
+
+#### Scenario: API key is required
+- **WHEN** the LLM provider needs an OpenAI API key
+- **THEN** the key SHALL be loaded from environment or local uncommitted configuration and SHALL NOT be committed
+
+#### Scenario: Audio is processed
+- **WHEN** user speech is captured for STT
+- **THEN** raw audio SHALL NOT be persisted by default
+
+#### Scenario: Cloud LLM is invoked
+- **WHEN** the system sends data to the cloud LLM
+- **THEN** it SHALL send transcript text and necessary conversation context, not raw microphone audio
+
+### Requirement: Testability
+The system SHALL be designed so each stage can be tested with mock providers and file-based audio fixtures.
+
+#### Scenario: Pipeline is unit tested
+- **WHEN** mock providers produce deterministic events
+- **THEN** tests SHALL verify state transitions, context updates, and error recovery
+
+#### Scenario: Audio fixture is replayed
+- **WHEN** a file-based audio fixture containing “小杰小杰” and user speech is replayed in a test Transport
+- **THEN** the pipeline SHALL be testable without using a live microphone
+
+#### Scenario: OpenSpec planning validation runs
+- **WHEN** this planning change is complete
+- **THEN** `openspec validate add-voice-pet-pipeline --strict` and `openspec validate --all --strict` SHALL pass
+
+### Requirement: Module-level git commit gates
+The system implementation process SHALL require an immediate Git commit after each major module or milestone is completed, where a major module means a top-level task group from `tasks.md` or a phase milestone from `proposal.md`.
+
+#### Scenario: Major module is completed
+- **WHEN** an implementer completes a top-level task group or milestone phase
+- **THEN** the implementer SHALL run the applicable build, tests, lint/type checks, and OpenSpec validation before committing
+
+#### Scenario: Build validation fails before commit
+- **WHEN** the applicable build, tests, lint/type checks, or OpenSpec validation fail
+- **THEN** the implementer SHALL NOT create the module commit until the failure is fixed and validation passes
+
+#### Scenario: Module commit is created
+- **WHEN** validation passes for the completed module
+- **THEN** the implementer SHALL immediately create a Git commit using the Chinese message format “[模块名]：完成[具体功能描述]，包含[关键变更]”
+
+#### Scenario: Intermediate work remains after module completion
+- **WHEN** module-related changes remain unstaged or uncommitted after the module is completed
+- **THEN** the implementer SHALL include those changes in the module commit or explicitly separate unrelated external changes before continuing to the next module
+
@@ -2,13 +2,33 @@ from __future__ import annotations

 import argparse
 import json
+import re
+import subprocess
+from pathlib import Path

+from .assets import validate_pet_assets
 from .config import AppConfig
+from .conversation import ConversationContext
+from .llm import MockLlmProvider, OpenAICompatibleLlmProvider
+from .models import AudioFrame, ProviderError
+from .pipeline import VoicePipeline
+from .stt import MetadataSttProvider
+from .transport import MemoryAudioTransport
+from .tts import SineTtsProvider
+from .vad import EnergyVadProvider, VadRecorder
+from .wakeword import KeywordWakeWordProvider


 def main(argv: list[str] | None = None) -> int:
    parser = argparse.ArgumentParser(prog="owner-voice-pet")
    parser.add_argument("--show-config", action="store_true", help="Print non-secret config summary")
+    subparsers = parser.add_subparsers(dest="command")
+    subparsers.add_parser("acceptance", help="Run deterministic end-to-end pipeline acceptance")
+    subparsers.add_parser("validate-assets", help="Validate project pet assets")
+    subparsers.add_parser("security-check", help="Scan tracked files for leaked API keys")
+    smoke = subparsers.add_parser("llm-smoke", help="Call configured OpenAI/NewAPI endpoint")
+    smoke.add_argument("--message", default="用一句中文回复：小杰在线。")
+    smoke.add_argument("--no-stream", action="store_true")
    args = parser.parse_args(argv)

    if args.show_config:
@@ -22,6 +42,7 @@ def main(argv: list[str] | None = None) -> int:
                    "llm_base_url": config.llm_base_url,
                    "llm_model": config.llm_model,
                    "llm_api_style": config.llm_api_style,
+                    "llm_stream": config.llm_stream,
                    "llm_api_key_present": bool(config.llm_api_key),
                    "asset_dir": str(config.asset_dir),
                },
@@ -31,5 +52,99 @@ def main(argv: list[str] | None = None) -> int:
        )
        return 0

+    if args.command == "validate-assets":
+        infos = validate_pet_assets(AppConfig.from_env().asset_dir)
+        print(json.dumps({"assets": len(infos), "valid": True}, ensure_ascii=False, sort_keys=True))
+        return 0
+
+    if args.command == "security-check":
+        leaks = find_secret_leaks(Path.cwd())
+        print(json.dumps({"secret_leaks": leaks, "valid": not leaks}, ensure_ascii=False, sort_keys=True))
+        return 1 if leaks else 0
+
+    if args.command == "acceptance":
+        result = run_acceptance()
+        print(json.dumps(result, ensure_ascii=False, sort_keys=True))
+        return 0 if result["success"] else 1
+
+    if args.command == "llm-smoke":
+        config = AppConfig.from_env()
+        if args.no_stream:
+            config = AppConfig(
+                wake_word=config.wake_word,
+                sample_rate=config.sample_rate,
+                channels=config.channels,
+                llm_base_url=config.llm_base_url,
+                llm_api_key=config.llm_api_key,
+                llm_model=config.llm_model,
+                llm_api_style=config.llm_api_style,
+                llm_stream=False,
+                audio_input_device=config.audio_input_device,
+                audio_output_device=config.audio_output_device,
+                asset_dir=config.asset_dir,
+                log_dir=config.log_dir,
+                context_max_messages=config.context_max_messages,
+                context_max_chars=config.context_max_chars,
+            )
+        try:
+            provider = OpenAICompatibleLlmProvider(config, timeout_s=30)
+            messages = [ConversationContext().build_llm_messages()[0]]
+            messages.append(__import__("owner_voice_pet.models", fromlist=["Message"]).Message("user", args.message, 1.0))
+            text = "".join(delta.text_delta for delta in provider.stream_reply(messages)).strip()
+            print(json.dumps({"ok": bool(text), "reply_preview": text[:80]}, ensure_ascii=False, sort_keys=True))
+            return 0 if text else 1
+        except ProviderError as exc:
+            print(json.dumps({"ok": False, "code": exc.code.value, "message": exc.message}, ensure_ascii=False, sort_keys=True))
+            return 1
+
    parser.print_help()
    return 0
+
+
+def run_acceptance() -> dict[str, object]:
+    validate_pet_assets(AppConfig.from_env().asset_dir)
+    frames = [
+        AudioFrame(b"\x80\x80", 16000, 1, 0, 0, {"duration_ms": 20, "wake_word": "小杰小杰", "wake_confidence": 0.99}),
+        AudioFrame(b"\xff\xff", 16000, 1, 20, 1, {"duration_ms": 20, "speech": True, "transcript": "你是谁"}),
+        AudioFrame(b"\xff\xff", 16000, 1, 40, 2, {"duration_ms": 20, "speech": True}),
+        AudioFrame(b"\x80\x80", 16000, 1, 60, 3, {"duration_ms": 20, "speech": False}),
+        AudioFrame(b"\x80\x80", 16000, 1, 80, 4, {"duration_ms": 20, "speech": False}),
+    ]
+    transport = MemoryAudioTransport(frames)
+    pipeline = VoicePipeline(
+        transport=transport,
+        wakeword=KeywordWakeWordProvider(),
+        vad_recorder=VadRecorder(EnergyVadProvider(), min_duration_ms=40, end_silence_ms=40),
+        stt=MetadataSttProvider(),
+        context=ConversationContext(),
+        llm=MockLlmProvider(["我是小杰桌宠，已经在线。"]),
+        tts=SineTtsProvider(),
+    )
+    pipeline.load()
+    result = pipeline.run_once()
+    return {
+        "success": result.success,
+        "transcript": result.transcript,
+        "assistant_text": result.assistant_text,
+        "played_segments": result.played_segments,
+        "states": [state.value for state in result.states],
+    }
+
+
+def find_secret_leaks(root: Path) -> list[str]:
+    completed = subprocess.run(["git", "ls-files"], cwd=root, check=True, stdout=subprocess.PIPE, text=True)
+    pattern = re.compile(r"sk-[A-Za-z0-9_\\-]{16,}")
+    leaks: list[str] = []
+    for rel in completed.stdout.splitlines():
+        path = root / rel
+        if not path.exists():
+            continue
+        if path.suffix.lower() in {".png", ".jpg", ".jpeg", ".webp"}:
+            continue
+        try:
+            text = path.read_text(encoding="utf-8")
+        except UnicodeDecodeError:
+            continue
+        if pattern.search(text):
+            leaks.append(rel)
+    return leaks
@@ -12,10 +12,11 @@ class AppConfig:
    wake_word: str = "小杰小杰"
    sample_rate: int = 16000
    channels: int = 1
-    llm_base_url: str = "https://api.openai.com"
+    llm_base_url: str = "https://newapi.mkbk.shop"
    llm_api_key: str | None = None
-    llm_model: str = "gpt-4o-mini"
+    llm_model: str = "gpt-5.4-mini"
    llm_api_style: str = "chat_completions"
+    llm_stream: bool = True
    audio_input_device: str | None = None
    audio_output_device: str | None = None
    asset_dir: Path = Path("assets/pet")
@@ -33,10 +34,11 @@ class AppConfig:
            wake_word=get("WAKE_WORD", "小杰小杰") or "小杰小杰",
            sample_rate=int(get("SAMPLE_RATE", "16000") or "16000"),
            channels=int(get("CHANNELS", "1") or "1"),
-            llm_base_url=(get("LLM_BASE_URL", "https://api.openai.com") or "").rstrip("/"),
+            llm_base_url=(get("LLM_BASE_URL", "https://newapi.mkbk.shop") or "").rstrip("/"),
            llm_api_key=get("LLM_API_KEY"),
-            llm_model=get("LLM_MODEL", "gpt-4o-mini") or "gpt-4o-mini",
+            llm_model=get("LLM_MODEL", "gpt-5.4-mini") or "gpt-5.4-mini",
            llm_api_style=get("LLM_API_STYLE", "chat_completions") or "chat_completions",
+            llm_stream=(get("LLM_STREAM", "1") or "1").lower() not in {"0", "false", "no"},
            audio_input_device=get("AUDIO_INPUT_DEVICE"),
            audio_output_device=get("AUDIO_OUTPUT_DEVICE"),
            asset_dir=Path(get("ASSET_DIR", "assets/pet") or "assets/pet"),
@@ -87,4 +89,14 @@ class AppConfig:
                    "startup",
                )
            )
+        if not self.llm_base_url.startswith(("http://", "https://")):
+            errors.append(
+                ProviderError(
+                    ErrorCode.CONFIG_MISSING_VALUE,
+                    "OWNER_LLM_BASE_URL must start with http:// or https://",
+                    False,
+                    "config",
+                    "startup",
+                )
+            )
        return errors
@@ -48,7 +48,7 @@ class OpenAICompatibleLlmProvider:
        payload = {
            "model": self.config.llm_model,
            "messages": [{"role": m.role, "content": m.content} for m in messages],
-            "stream": True,
+            "stream": self.config.llm_stream,
        }
        yield from self._post_stream(
            f"{self.config.llm_base_url}/v1/chat/completions",
@@ -60,7 +60,7 @@ class OpenAICompatibleLlmProvider:
        payload = {
            "model": self.config.llm_model,
            "input": [{"role": m.role, "content": m.content} for m in messages],
-            "stream": True,
+            "stream": self.config.llm_stream,
        }
        yield from self._post_stream(
            f"{self.config.llm_base_url}/v1/responses",
@@ -86,8 +86,10 @@ class OpenAICompatibleLlmProvider:
        )
        try:
            with self.urlopen(request, timeout=self.timeout_s) as response:
+                raw_body = _read_response_body(response)
                emitted = False
-                for raw_line in response:
+                if raw_body.lstrip().startswith(b"data:"):
+                    for raw_line in raw_body.splitlines():
                        line = raw_line.decode("utf-8", errors="replace").strip()
                        if not line or not line.startswith("data:"):
                            continue
@@ -100,6 +102,10 @@ class OpenAICompatibleLlmProvider:
                        if delta is not None:
                            emitted = True
                            yield delta
+                else:
+                    for delta in _parse_non_stream_json(json.loads(raw_body.decode("utf-8"))):
+                        emitted = True
+                        yield delta
                if not emitted:
                    raise ProviderError(
                        ErrorCode.LLM_EMPTY_REPLY,
@@ -145,5 +151,36 @@ def _parse_responses_sse(event: dict[str, Any]) -> ReplyDelta | None:
    return None


+def _parse_non_stream_json(event: dict[str, Any]) -> Iterable[ReplyDelta]:
+    if "choices" in event:
+        choices = event.get("choices") or []
+        if choices:
+            message = choices[0].get("message") or {}
+            text = str(message.get("content") or "")
+            if text:
+                yield ReplyDelta(text, is_sentence_boundary=_ends_sentence(text), finish_reason="stop")
+        return
+    if "output_text" in event:
+        text = str(event.get("output_text") or "")
+        if text:
+            yield ReplyDelta(text, is_sentence_boundary=_ends_sentence(text), finish_reason="stop")
+        return
+    output = event.get("output") or []
+    parts: list[str] = []
+    for item in output:
+        for content in item.get("content", []):
+            if content.get("type") in {"output_text", "text"}:
+                parts.append(str(content.get("text") or ""))
+    text = "".join(parts)
+    if text:
+        yield ReplyDelta(text, is_sentence_boundary=_ends_sentence(text), finish_reason="stop")
+
+
+def _read_response_body(response: Any) -> bytes:
+    if hasattr(response, "read"):
+        return response.read()
+    return b"".join(response)
+
+
 def _ends_sentence(text: str) -> bool:
    return text.endswith(("。", "！", "？", ".", "!", "?", "\n"))
@@ -0,0 +1,37 @@
+from __future__ import annotations
+
+import io
+import json
+import unittest
+from contextlib import redirect_stdout
+
+from owner_voice_pet.cli import main
+
+
+class CliAcceptanceTests(unittest.TestCase):
+    def call(self, *args: str) -> tuple[int, dict[str, object]]:
+        buffer = io.StringIO()
+        with redirect_stdout(buffer):
+            code = main(list(args))
+        return code, json.loads(buffer.getvalue())
+
+    def test_acceptance_command_runs_end_to_end(self) -> None:
+        code, data = self.call("acceptance")
+        self.assertEqual(code, 0)
+        self.assertTrue(data["success"])
+        self.assertEqual(data["transcript"], "你是谁")
+        self.assertGreaterEqual(data["played_segments"], 1)
+
+    def test_validate_assets_command(self) -> None:
+        code, data = self.call("validate-assets")
+        self.assertEqual(code, 0)
+        self.assertTrue(data["valid"])
+
+    def test_security_check_command_has_no_leaks(self) -> None:
+        code, data = self.call("security-check")
+        self.assertEqual(code, 0)
+        self.assertEqual(data["secret_leaks"], [])
+
+
+if __name__ == "__main__":
+    unittest.main()
@@ -62,6 +62,7 @@ class ModelsConfigTests(unittest.TestCase):
            self.assertEqual(config.llm_base_url, "https://newapi.mkbk.shop")
            self.assertEqual(config.llm_api_key, "secret-value")
            self.assertEqual(config.llm_model, "test-model")
+            self.assertTrue(config.llm_stream)
            self.assertEqual(config.validate_basic(), [])
        finally:
            os.environ.clear()
@@ -120,12 +120,14 @@ class PipelineLlmTtsTests(unittest.TestCase):
            def __exit__(self, *args: object) -> None:
                return None

-            def __iter__(self):
+            def read(self):
                event = {"choices": [{"delta": {"content": "你好。"}, "finish_reason": None}]}
                done = {"choices": [{"delta": {}, "finish_reason": "stop"}]}
-                yield f"data: {json.dumps(event, ensure_ascii=False)}\n".encode()
-                yield f"data: {json.dumps(done, ensure_ascii=False)}\n".encode()
-                yield b"data: [DONE]\n"
+                return (
+                    f"data: {json.dumps(event, ensure_ascii=False)}\n"
+                    f"data: {json.dumps(done, ensure_ascii=False)}\n"
+                    "data: [DONE]\n"
+                ).encode()

        requests = []

@@ -146,6 +148,27 @@ class PipelineLlmTtsTests(unittest.TestCase):
        self.assertTrue(body["stream"])
        self.assertEqual(body["model"], "test-model")

+    def test_openai_non_stream_json_parser(self) -> None:
+        class FakeResponse:
+            def __enter__(self) -> "FakeResponse":
+                return self
+
+            def __exit__(self, *args: object) -> None:
+                return None
+
+            def read(self):
+                return json.dumps({"choices": [{"message": {"content": "非流式回复。"}}]}).encode()
+
+        config = AppConfig(
+            llm_base_url="https://newapi.mkbk.shop",
+            llm_api_key="secret",
+            llm_model="test-model",
+            llm_api_style="chat_completions",
+            llm_stream=False,
+        )
+        provider = OpenAICompatibleLlmProvider(config, urlopen=lambda request, timeout: FakeResponse())
+        self.assertEqual(list(provider.stream_reply([Message("user", "你好", 1.0)]))[0].text_delta, "非流式回复。")
+

 if __name__ == "__main__":
    unittest.main()