jiezeng2004-design/dsh-requirements-alignment

jiezeng2004-design★ 3TypeScriptLast synced: 2026-08-14

Open on GitHub

Lightweight requirement alignment for DeepSeek Harness — align important decisions before execution without a full spec workflow.

README excerpt

dsh-requirements-alignment Lightweight requirement alignment for DeepSeek Harness. Overview dsh-requirements-alignment adds a lightweight alignment layer before DeepSeek Harness agents execute ambiguous or high-impact requests. It asks only when a decision materially affects direction, remembers resolved choices within the session, and stays out of the way once the intent is clear. You decide the direction. The agent decides the engineering. Why this plugin exists An agent can produce technically sound work in the wrong product direction when a request leaves a high-impact choice unresolved. This plugin applies the minimum intervention needed to surface that choice before execution: one focused question at a time, followed by autonomous implementation once the direction is coherent. It re-aligns only when the task direction materially changes. Focused scope dsh-requirements-alignment intentionally focuses only on requirement direction alignment. It does not impose a complete engineering workflow and it does not change the user's normal DSH workflow. It is not : - a spec, TDD, code-review, or delivery-verification framework; - a complete software development methodology; - an interv…

View full README on GitHub →
Agentsagentai-agentcordisdeepseek-harnessdsh-pluginrequirement-alignment

Category