supart/DSH-FHL-Image-Plugin ↗★ 2
@fhl-plugins/dsh-fhl-image
提供FHL图像生成、编辑及凭据配置工具 适合需要在会话中配置专用API并进行AI图像生成与编辑的用户。
安裝
npx -p @deepseek-ai/dsh dsh plugin --profile web add github:supart/DSH-FHL-Image-Plugin說明文件
閱讀完整 README ↗聊天配置生图 API
新建会话,先确认工具列表中有上面三个工具,然后发送:
我想配置 FHL 生图专用 API。请使用 fhl_image_configure 工具,并先告诉我需要如何提供一个或多个 Worker Key;不要调用生图工具。
Agent 询问后,把一个或多个 Worker Key 粘贴到聊天框,每个 Key 一行。配置阶段只应
调用 fhl_image_configure,成功后只根据脱敏的 Worker 数量确认配置。
聊天配置有一个重要边界:原始 Key 会随用户消息经过当前模型,并可能保留在 DSH 会话
历史中。不能接受这种传输方式的凭据,应改用 DSH 自身的凭据机制直接写入对应引用名
(FHL_IMAGE_API_KEY、FHL_IMAGE_API_KEY_2……)。插件配置项 apiKeyEnv 只是这些
引用名的前缀,不是环境变量。不要把 Key 发到 GitHub、Codex 主对话、截图、日志
或 Issue。
覆盖插件配置
默认值写在 bundle 层里,不读取环境变量。需要修改时,在 profile 补丁层覆盖:
## Chat Configuration
```text
I want to configure FHL image workers. Use fhl_image_configure first; do not call image tools.
For multiple workers, send one key per line. The Agent calls
fhl_image_configure, which stores the keys through the DSH credential provider
and returns only a masked worker count.
This mode is intentionally chat-based, so the original key is part of the user
message sent to the selected model and may be retained in the DSH session
history. A key that must never reach the model provider should be written into
the credential store directly under the reference name
(FHL_IMAGE_API_KEY, FHL_IMAGE_API_KEY_2, …). The apiKeyEnv config value is
only the prefix of those reference names; it is not an environment variable.
The plugin never includes key values in tool results, image artifacts, normal
diagnostics, or its own error messages.
Real Usage Examples

