subagent-driven-development

Automating implementation plans prevents context contamination by dispatching a fresh subagent for each independent task, ensuring isolated context for every step. This approach enforces two-stage reviews for spec compliance and code quality before integration.

Executing modular tasks concurrently requires dispatching fresh subagents to handle multiple independent tasks in parallel. This method automatically enforces a two-stage review process, checking both spec compliance and code quality for each delegated task.

Running a feature plan requires invoking the subagent-driven-development skill and pointing it to the plan file path, such as docs/plans/feature-plan.md. The skill then reads the plan and dispatches fresh subagents to execute each independent task.

Subagent-driven development is designed for software projects where a detailed plan is broken into small, independent tasks. Without modular task breakdowns, the skill cannot effectively dispatch isolated subagents or enforce its two-stage spec compliance and code quality reviews.

Manually executing implementation plans leads to inconsistent code quality because of context contamination and missed spec details. Automating this process with fresh subagents enforces isolated contexts and automatic code quality assessments for every task.

Using this skill requires the ability to invoke delegate_task, manage todo lists, and perform file and terminal toolset operations. Without the delegate_task toolset, the skill cannot dispatch fresh subagents or handle parallel execution of independent tasks.