dsh-goal
Run the following command in DeepSeek Harness:
dsh plugin install GooDAnDReaDY/dsh-goal
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install GooDAnDReaDY/dsh-goal in your DeepSeek Harness terminal to install the plugin, and visit https://github.com/GooDAnDReaDY/dsh-goal for full documentation.
About this plugin
Complex engineering tasks routinely exceed what a single conversational turn can handle. Traditional agent interactions stumble under the weight: context drifts away between turns, passive loops idle without forward motion, and users are forced to repeatedly hand-prompt the process forward. dsh-goal introduces a full autonomous goal execution engine to DeepSeek Harness, decomposing high-level intent into sequenced milestones that the agent drives forward one by one without requiring manual re-engagement at every step.
The core experience centers on a persistent sticky top banner: a live elapsed timer, a status badge, and one-tap pause/resume controls are always in view. Under the hood, the engine provides a milestone state machine, ETA estimation based on completion velocity, token usage accumulation with Markdown export, a consecutive tool-failure circuit breaker, git checkpoint snapshots with one-click rollback, native browser push notifications, and Web Audio chimes for completion or failure. The agent interacts with the engine through three tool contracts — goal_set_milestones, goal_update_progress, and goal_finish — while also shadowing DSH built-in tools (update_goal, get_goal, create_goal) so that different model backends plug in without friction.
This plugin suits developers working complex engineering scenarios inside DeepSeek Harness — fixing bugs, YAGNI refactors, test coverage, code review, security audits, dependency upgrades, dead-code cleanup, performance tuning — anyone who wants the agent to autonomously run a full workflow while keeping progress transparent and interruption under their control.
Use Cases
- When the agent must autonomously drive multi-turn complex engineering tasks
- When milestone progress visualization with instant pause and resume is needed
- When token budget tracking and git checkpoint rollback are required for production work
Best For
- Developers doing complex engineering work inside DeepSeek Harness
- Users who need agents to run full workflows autonomously with transparent progress
- Heavy agent users who care about token budgets and safety circuit breakers
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.