Execute confirmed implementation plans by dispatching isolated subagents with rubber-duck verification.
npx skills add https://github.com/josipmusa/jds --skill jds-execute
Please help me install this Agent Skill. Skill: jds-execute Source: https://github.com/josipmusa/jds/tree/main/skills/jds-execute Command: npx skills add https://github.com/josipmusa/jds --skill jds-execute
Coordinates the execution of a confirmed implementation plan by dispatching isolated subagents, ensuring context isolation and a disciplined workflow.
Run jds-plan to generate a plan and then invoke jds-execute to run tasks one by one with isolated subagents.
High-intent search queries and answers about installing and using this skill.
To execute an implementation plan with isolated subagents, you need a confirmed plan stored under docs/jds/plans/ and then invoke the execution workflow to dispatch tasks one-by-one with context isolation and concurrency control.
Context isolation works by providing each dispatched subagent with only task-relevant context, preventing the leakage of previous session history and ensuring a disciplined workflow during plan execution.
The best way to verify task completion during plan execution is using a built-in rubber-duck reviewer workflow that validates results before marking individual tasks as complete.
Yes, you can control concurrency when executing implementation plans with multiple subagents through a configurable concurrency model that manages how isolated subagents process tasks sequentially or in parallel.
Yes, you need a confirmed implementation plan located under the docs/jds/plans/ directory before you can start dispatching isolated subagents to execute it.
Isolated subagents prevent session history leakage during workflow execution, ensuring each task receives only relevant context for disciplined context isolation and verifier-based completion validation.
tizeeGenerate implementation plans from feature specifications and save them to docs/plans.
cshyangRun the gsd-verifier agent on phase plans and task details.
badaniyaDispatch a fresh subagent per task and run two-stage reviews.
nesquikmVerify web UI layouts and rendering via rubber duck MCP with Chrome DevTools.
