Category: task
Model Studio Skills Minimal Test
Run minimal validation for currently available Model Studio skills in this repo and record results.
Prerequisites
- - Install SDK (virtual environment recommended to avoid PEP 668 restrictions):
CODEBLOCK0
- - Configure
DASHSCOPE_API_KEY (environment variable preferred; or dashscope_api_key in ~/.alibabacloud/credentials).
Test Matrix (currently supported)
1) Text-to-image → skills/ai/image/aliyun-qwen-image/
2) Image editing → skills/ai/image/aliyun-qwen-image-edit/
3) Text-to-video / Image-to-video (i2v) → skills/ai/video/aliyun-wan-video/
4) Reference-to-video (r2v) → skills/ai/video/aliyun-wan-r2v/
5) TTS → skills/ai/audio/aliyun-qwen-tts/
6) ASR transcription (non-realtime) → skills/ai/audio/aliyun-qwen-asr/
7) Realtime ASR → skills/ai/audio/aliyun-qwen-asr-realtime/
8) Realtime TTS → skills/ai/audio/aliyun-qwen-tts-realtime/
9) Live speech translation → skills/ai/audio/aliyun-qwen-livetranslate/
10) CosyVoice voice clone → skills/ai/audio/aliyun-cosyvoice-voice-clone/
11) CosyVoice voice design → skills/ai/audio/aliyun-cosyvoice-voice-design/
12) Voice clone → skills/ai/audio/aliyun-qwen-tts-voice-clone/
13) Voice design → skills/ai/audio/aliyun-qwen-tts-voice-design/
14) Omni multimodal → skills/ai/multimodal/aliyun-qwen-omni/
15) Visual reasoning → skills/ai/multimodal/aliyun-qvq/
16) Text embedding → skills/ai/search/aliyun-qwen-text-embedding/
17) Rerank → skills/ai/search/aliyun-qwen-rerank/
18) Video editing → INLINECODE20
If new capability tests are needed, create corresponding skill first (use skills/ai/misc/aliyun-modelstudio-crawl-and-skill/ to refresh model list).
Minimal Flow Per Capability
- 1. Open target sub-skill directory and read
SKILL.md. - Choose one minimal input example and recommended model.
- Run SDK call or script.
- Record model, request summary, response summary, duration, and status.
Result Template
Save as output/aliyun-modelstudio-entry-test-results.md:
CODEBLOCK1
Failure Handling
- - If parameters are unclear, check target sub-skill
SKILL.md or references/*.md. - If model is unavailable, refresh model list and retry.
- For auth issues, verify
DASHSCOPE_API_KEY (env var or ~/.alibabacloud/credentials).
Validation
CODEBLOCK2
Pass criteria: command exits 0 and output/aliyun-modelstudio-entry-test/validate.txt is generated.
Output And Evidence
- - Save artifacts, command outputs, and API response summaries under
output/aliyun-modelstudio-entry-test/. - Include key parameters (region/resource id/time range) in evidence files for reproducibility.
Workflow
1) Confirm user intent, region, identifiers, and whether the operation is read-only or mutating.
2) Run one minimal read-only query first to verify connectivity and permissions.
3) Execute the target operation with explicit parameters and bounded scope.
4) Verify results and save output/evidence files.
References
- - Source list: INLINECODE30
技能名称: aliyun-modelstudio-entry-test
详细描述:
类别: 任务
Model Studio 技能最小化测试
对该仓库中当前可用的 Model Studio 技能运行最小化验证并记录结果。
前提条件
- - 安装 SDK(建议使用虚拟环境以避免 PEP 668 限制):
bash
python3 -m venv .venv
. .venv/bin/activate
python -m pip install dashscope
- - 配置 DASHSCOPEAPIKEY(优先使用环境变量;或 ~/.alibabacloud/credentials 中的 dashscopeapikey)。
测试矩阵(当前支持)
1) 文生图 → skills/ai/image/aliyun-qwen-image/
2) 图像编辑 → skills/ai/image/aliyun-qwen-image-edit/
3) 文生视频 / 图生视频 (i2v) → skills/ai/video/aliyun-wan-video/
4) 参考视频生成 (r2v) → skills/ai/video/aliyun-wan-r2v/
5) 语音合成 (TTS) → skills/ai/audio/aliyun-qwen-tts/
6) 语音识别转录 (非实时) → skills/ai/audio/aliyun-qwen-asr/
7) 实时语音识别 → skills/ai/audio/aliyun-qwen-asr-realtime/
8) 实时语音合成 → skills/ai/audio/aliyun-qwen-tts-realtime/
9) 实时语音翻译 → skills/ai/audio/aliyun-qwen-livetranslate/
10) CosyVoice 声音克隆 → skills/ai/audio/aliyun-cosyvoice-voice-clone/
11) CosyVoice 声音设计 → skills/ai/audio/aliyun-cosyvoice-voice-design/
12) 声音克隆 → skills/ai/audio/aliyun-qwen-tts-voice-clone/
13) 声音设计 → skills/ai/audio/aliyun-qwen-tts-voice-design/
14) 全能多模态 → skills/ai/multimodal/aliyun-qwen-omni/
15) 视觉推理 → skills/ai/multimodal/aliyun-qvq/
16) 文本嵌入 → skills/ai/search/aliyun-qwen-text-embedding/
17) 重排序 → skills/ai/search/aliyun-qwen-rerank/
18) 视频编辑 → skills/ai/video/aliyun-wan-edit/
如果需要测试新能力,请先创建对应的技能(使用 skills/ai/misc/aliyun-modelstudio-crawl-and-skill/ 刷新模型列表)。
每个能力的最小化流程
- 1. 打开目标子技能目录并读取 SKILL.md。
- 选择一个最小输入示例和推荐模型。
- 运行 SDK 调用或脚本。
- 记录模型、请求摘要、响应摘要、耗时和状态。
结果模板
保存为 output/aliyun-modelstudio-entry-test-results.md:
Model Studio 技能测试结果
- - 日期:YYYY-MM-DD
- 环境:区域 / API_BASE / 认证方式
| 能力 | 子技能 | 模型 | 请求摘要 | 结果摘要 | 状态 | 备注 |
|---|
| 文生图 | aliyun-qwen-image | <模型ID> | ... | ... | 通过/失败 | ... |
| 图像编辑 |
aliyun-qwen-image-edit | <模型ID> | ... | ... | 通过/失败 | ... |
| 图生视频 (i2v) | aliyun-wan-video | <模型ID> | ... | ... | 通过/失败 | ... |
| 参考视频生成 (r2v) | aliyun-wan-r2v | <模型ID> | ... | ... | 通过/失败 | ... |
| 语音合成 | aliyun-qwen-tts | <模型ID> | ... | ... | 通过/失败 | ... |
| 语音识别 (非实时) | aliyun-qwen-asr | <模型ID> | ... | ... | 通过/失败 | ... |
| 实时语音识别 | aliyun-qwen-asr-realtime | <模型ID> | ... | ... | 通过/失败 | ... |
| 实时语音合成 | aliyun-qwen-tts-realtime | <模型ID> | ... | ... | 通过/失败 | ... |
| 实时语音翻译 | aliyun-qwen-livetranslate | <模型ID> | ... | ... | 通过/失败 | ... |
| CosyVoice 声音克隆 | aliyun-cosyvoice-voice-clone | <模型ID> | ... | ... | 通过/失败 | ... |
| CosyVoice 声音设计 | aliyun-cosyvoice-voice-design | <模型ID> | ... | ... | 通过/失败 | ... |
| 声音克隆 | aliyun-qwen-tts-voice-clone | <模型ID> | ... | ... | 通过/失败 | ... |
| 声音设计 | aliyun-qwen-tts-voice-design | <模型ID> | ... | ... | 通过/失败 | ... |
| 全能多模态 | aliyun-qwen-omni | <模型ID> | ... | ... | 通过/失败 | ... |
| 视觉推理 | aliyun-qvq | <模型ID> | ... | ... | 通过/失败 | ... |
| 文本嵌入 | aliyun-qwen-text-embedding | <模型ID> | ... | ... | 通过/失败 | ... |
| 重排序 | aliyun-qwen-rerank | <模型ID> | ... | ... | 通过/失败 | ... |
| 视频编辑 | aliyun-wan-edit | <模型ID> | ... | ... | 通过/失败 | ... |
失败处理
- - 如果参数不明确,请检查目标子技能的 SKILL.md 或 references/*.md。
- 如果模型不可用,请刷新模型列表并重试。
- 对于认证问题,请验证 DASHSCOPEAPIKEY(环境变量或 ~/.alibabacloud/credentials)。
验证
bash
mkdir -p output/aliyun-modelstudio-entry-test
echo validation_placeholder > output/aliyun-modelstudio-entry-test/validate.txt
通过标准:命令退出码为 0 且 output/aliyun-modelstudio-entry-test/validate.txt 已生成。
输出与证据
- - 将产物、命令输出和 API 响应摘要保存到 output/aliyun-modelstudio-entry-test/ 下。
- 在证据文件中包含关键参数(区域/资源ID/时间范围)以确保可复现性。
工作流程
1) 确认用户意图、区域、标识符以及操作是只读还是修改性。
2) 首先运行一个最小化的只读查询以验证连接和权限。
3) 使用明确的参数和有限的范围执行目标操作。
4) 验证结果并保存输出/证据文件。
参考资料
- - 来源列表:references/sources.md