AI Agent Hub
Back to plugins
🖥️

dsh-ask-first

Client Updated 2026.09.05

Run the following command in DeepSeek Harness:

dsh plugin install 1052326311/dsh-ask-first

Paste the following prompt into your AI chat to install this plugin:

Run dsh plugin install 1052326311/dsh-ask-first inside DeepSeek Harness to install the plugin; the source repository is at https://github.com/1052326311/dsh-ask-first.

About this plugin

When you tell an agent "build me an admin dashboard," it often rushes into code, silently deciding user roles, permissions, information hierarchy, and success criteria on the fly. The result works, but it is rarely what you intended. dsh-ask-first closes that gap: before any substantive work begins, it turns a fuzzy request into a confirmed plan.

The plugin follows an inspect-then-ask loop. The agent first reads the repository and current product for evidence, then asks only about decisions that the evidence cannot settle and that would materially change the design. Each round carries at most three questions, and earlier answers shape later ones; there is no fixed questionnaire. Once discovery wraps, the agent drafts a concise brief covering outcome, users, scope, constraints, and acceptance checks, and you confirm or revise it in the native DSH dialog. Precise, low-risk tasks like fixing a named typo skip the interview entirely.

It is built for developers working on product-level or cross-module features inside DSH where the requirement boundary is still open. The plugin adds no second chat UI, no external service, and no database; it layers on the native ask_user_question and Plan Mode primitives, ready to use the moment you install it.

Use Cases

  • Before a vague product request, let the agent confirm key decisions before coding
  • During cross-module development, lock down roles, permissions, and acceptance checks through progressive questions
  • Under Plan Mode, unify requirement confirmation with implementation approval to avoid duplicate dialogs

Best For

  • Developers building product-level or cross-module features inside DSH
  • Teams that want agents to align on requirements before writing code
  • Solo developers seeking a controllable, auditable discovery workflow