governed-workflow-for-dsh
Run the following command in DeepSeek Harness:
dsh plugin install zcx369658780/governed-workflow-for-dsh
Paste the following prompt into your AI chat to install this plugin:
In DeepSeek Harness, you can install this plugin using the command `dsh plugin install zcx369658780/governed-workflow-for-dsh`, with the full source code at https://github.com/zcx369658780/governed-workflow-for-dsh.
About this plugin
In vibe-coding projects, we often face a challenge: once given a prompt, coding agents modify code freely, leading to scope drift, accidental changes, and a lack of clear review boundaries. The dsh-governed-workflow plugin addresses this by introducing a governed sequence, using GitHub Issues as the authority source for tasks. The lifecycle state machine ensures that protected mutations only occur during the RUNNING state, and modifications freeze upon completion or blocking before moving to independent review, reducing drift and unintended edits in long sessions.
Key capabilities include: GitHub Issue-based task authority, strict lifecycle state management (e.g., from AUTHORITY_OBSERVED to REVIEW_PENDING), guarding modifications to tools like bash, write, and edit only when in RUNNING state, providing read-only governance_status and allowlisted governance_transition actions, and terminal freeze in states like BLOCKED or COMPLETED. All authority observations and transitions are logged as bounded governance evidence for review and replay, enhancing traceability and discipline.
This plugin is best suited for medium-to-large, multi-file, long-running vibe-coding projects where agents execute commands or modify real code and mistakes carry high costs—such as apps, backend services, research tooling, or maintained repositories. It is less ideal for one-off scripts or tiny throwaway prototypes, where governance overhead may outweigh the benefits. By trading some speed for structure, it offers clearer boundaries and better support for projects requiring real review, rollback, or release processes.
Screenshots
Use Cases
- Multi-stage software development projects requiring clear task authority and independent review.
- Preventing scope drift and accidental edits during long coding sessions.
- High-stakes vibe-coding projects involving real code modifications and deployments.
Best For
- Teams developing medium-to-large applications or backend services.
- Project maintainers needing code review and rollback mechanisms.
- Engineers working on research tooling or automation systems.
Related Plugins
A method pack that makes AI coding agents plan against your real baseline, prove completion with fresh evidence, and reduce reworks and unsafe changes.
Turns the DeepSeek Harness session into a captain that builds a durable sub-agent team, splits goals into dependency-aware tasks, and coordinates work via direct messages and a live Web UI.
Gives coding agents design judgment, letting Claude Code, Cursor, and 70+ agents generate and iterate high-quality UI, presentations, and graphics right from the terminal.
Run the Pi ecosystem's plugins on DeepSeek Harness, unmodified, via a compatibility layer that implements Pi's public extension ABI on DSH's native services.