dsh-prompt-injector

Generic per-turn prompt injector for the DeepSeek Harness web profile — manage a list of context prompts (with enable switch) in the settings page; every round of conversation injects each enabled prompt as a notice line (「上下文注入 xxx」) into the model context, exactly like the mem0 recall reminder. Host-only logic + client settings card; zero dsh source changes.

AI 分析

核心用途是允许用户在设置中管理一组提示词，并在每轮对话中将启用的提示词作为上下文提示行自动注入模型。适合需要全局或高频向模型灌输特定行为准则的用户。

memorynotifications

パッケージ: dsh-prompt-injector
バージョン: 0.1.0
ライセンス: MIT
最終更新: 2026/08/27

インストール

$npx -p @deepseek-ai/dsh dsh plugin --profile web add github:runfali/dsh-prompt-injector

ドキュメント

README 全文を読む ↗

Configuration

Settings card — master switch, trivial-round filter, editable prompt list

Key	Type	Default	Meaning
`enabled`	boolean	`true`	Master switch.
`skipTrivial`	boolean	`true`	Skip trivial rounds (ack/greeting/continuation).
`prompts`	array	built-in default	Prompt list: `[{ id, title, text, enabled }]`.

The settings page edits these; the user layer wins over the composition defaults. To override via the profile patch: