refactor(jiume): import codex pet avatars

Author: luoyige

README.md

@@ -55,13 +55,10 @@ What should happen:
 3. The JiuMe desktop avatar appears.
 4. You can click the avatar and send a one-line task.
 
-No API key is required for the first local run. If you later want generated avatars or hosted model providers, add credentials in Setup or export:
+No API key is required for the first local run. Avatar art comes from a Codex pet package. Generate one with the local `hatch-pet` skill or download an existing Codex pet project, then import the folder or ZIP in Setup:
 
 ```bash
-export JIUME_OPENAI_API_KEY="..."
-export JIUME_OPENAI_BASE_URL="https://api.openai.com/v1"
-export JIUME_IMAGE_MODEL="gpt-image-2"
-export JIUME_IMAGE_PROVIDER="auto"
+.venv/bin/jiume-setup --pet ~/.codex/pets/my-pet --enable
 ```
 
 For a UI-only smoke test:

README_CN.md

@@ -55,13 +55,10 @@ uv pip install -e ".[test]"
 3. JiuMe 桌面头像出现。
 4. 你可以单击头像并发送一句话任务。
 
-第一次本地运行不需要 API Key。之后如果想接入生成头像或托管模型，可以在 Setup 里填写，也可以导出环境变量：
+第一次本地运行不需要 API Key。头像资产来自 Codex pet 包：用本机 `hatch-pet` skill 生成，或下载现成 Codex pet 项目，然后在 Setup 里导入目录或 ZIP：
 
 ```bash
-export JIUME_OPENAI_API_KEY="..."
-export JIUME_OPENAI_BASE_URL="https://api.openai.com/v1"
-export JIUME_IMAGE_MODEL="gpt-image-2"
-export JIUME_IMAGE_PROVIDER="auto"
+.venv/bin/jiume-setup --pet ~/.codex/pets/my-pet --enable
 ```
 
 如果只想离线查看头像 UI：

docs/en/JiuMe.md

@@ -69,18 +69,15 @@ Manually set the avatar state while testing:
 .venv/bin/jiume-state sleep
 ```
 
-## Optional Image Provider
+## Avatar Pet Package
 
-Image generation is optional. Without credentials, JiuMe still opens with local avatar assets.
+JiuMe does not generate avatar art in-app. Generate a Codex pet with `hatch-pet` or download an existing Codex pet package, then import its folder or ZIP:
 
 ```bash
-export JIUME_OPENAI_API_KEY="..."
-export JIUME_OPENAI_BASE_URL="https://api.openai.com/v1"
-export JIUME_IMAGE_MODEL="gpt-image-2"
-export JIUME_IMAGE_PROVIDER="auto"
+.venv/bin/jiume-setup --pet ~/.codex/pets/my-pet --enable
 ```
 
-Setup can persist these values to `~/.jiuwenswarm/jiume/config.env`.
+The package must contain `pet.json` and `spritesheet.webp`.
 
 ## Local Data
 

docs/zh/JiuMe.md

@@ -69,18 +69,15 @@ JiuMe 会把 Agent 过程映射成简单状态：idle、thinking、working、wai
 .venv/bin/jiume-state sleep
 ```
 
-## 可选图片服务
+## 头像 Pet 包
 
-图片生成是可选能力。没有配置凭证时，JiuMe 仍会使用本地头像资产打开。
+JiuMe 不再在应用内生成头像。请用 `hatch-pet` 生成 Codex pet，或下载现成 Codex pet 包，然后导入目录或 ZIP：
 
 ```bash
-export JIUME_OPENAI_API_KEY="..."
-export JIUME_OPENAI_BASE_URL="https://api.openai.com/v1"
-export JIUME_IMAGE_MODEL="gpt-image-2"
-export JIUME_IMAGE_PROVIDER="auto"
+.venv/bin/jiume-setup --pet ~/.codex/pets/my-pet --enable
 ```
 
-Setup 也可以把这些值保存到 `~/.jiuwenswarm/jiume/config.env`。
+包内需要包含 `pet.json` 和 `spritesheet.webp`。
 
 ## 本地数据
 

jiume/api/web_handlers.py

@@ -53,6 +53,9 @@ async def _handler(ws: Any, req_id: str, params: Any, session_id: str) -> None:
     def _job(job: DistillationJob) -> dict[str, Any]:
         return job.to_dict()
 
+    def _avatar_generation_removed(message: str) -> dict[str, Any]:
+        raise ValueError(message)
+
     channel.register_method(
         "twin.list",
         _wrap(lambda p: {"twins": store.list_twins(), "activeTwinId": store.get_active_twin_id()}),
@@ -68,16 +71,29 @@ def _job(job: DistillationJob) -> dict[str, Any]:
 
     channel.register_method(
         "twin.avatar.upload_source",
-        _wrap(lambda p: {"upload": avatars.upload_source(_twin_id(p, store), p)}),
+        _wrap(lambda _p: _avatar_generation_removed("photo avatar generation was removed; import a Codex pet package")),
     )
     channel.register_method(
         "twin.avatar.generate",
-        _wrap(lambda p: {"job": avatars.generate_avatar(_twin_id(p, store), str(p.get("provider") or "auto"))}),
+        _wrap(lambda _p: _avatar_generation_removed("avatar generation was removed; import a Codex pet package")),
     )
     channel.register_method(
         "twin.avatar.job_get",
         _wrap(lambda p: {"job": avatars.get_job(_twin_id(p, store), str(p.get("job_id") or p.get("jobId") or ""))}),
     )
+    channel.register_method("twin.avatar.list_pets", _wrap(lambda _p: {"pets": avatars.list_pets()}))
+    channel.register_method(
+        "twin.avatar.import_pet",
+        _wrap(
+            lambda p: {
+                "job": avatars.import_pet(
+                    _twin_id(p, store),
+                    pet_id=str(p.get("petId") or p.get("pet_id") or "").strip() or None,
+                    source_path=str(p.get("sourcePath") or p.get("source_path") or "").strip() or None,
+                )
+            }
+        ),
+    )
     channel.register_method(
         "twin.avatar.get_manifest",
         _wrap(lambda p: {"manifest": avatars.get_manifest(_twin_id(p, store))}),