vtm-expert

The vtm-expert skill acts as a specialized agent to manage the Virtual Task Manager (VTM) CLI, specifically designed to eliminate context bloat in AI development workflows. By providing surgical access to task manifests, the skill enables developers and Claude to interact with complex projects while maintaining a 99% reduction in token consumption. It serves as an intelligent bridge between architectural planning and code implementation, ensuring that state tracking, dependency resolution, and TDD-based test strategies are consistently applied across the development lifecycle.

• Intelligent command suggestions: Proactively identifies task-related user intent (e.g., 'What should I work on?') and suggests appropriate /vtm: commands.

• Surgical context loading: Retrieves only the necessary task metadata rather than full manifests, optimizing token usage for large-scale projects.

• Lifecycle management: Orchestrates task progression from 'Next' and 'Context' to 'Start' and 'Complete', ensuring atomic updates with automatic stats recalculation.

• Dependency and TDD enforcement: Validates that task prerequisites are met and adheres to specified test strategies before code implementation.

• Seamless CLI integration: Wraps the underlying vtm CLI for atomic file operations and maintains sync with project-level vtm.json files.

• Always run /vtm:next to verify work readiness before initiating tasks.

• Utilize /vtm:context prior to starting implementation to populate Claude with minimal, high-signal task data.

• Integrate with git hooks (pre-commit, post-checkout) provided in the plugin to maintain persistent data integrity and cache consistency.

• When batch planning, utilize the bridge commands like /plan:to-vtm to transform ADRs and technical specifications into actionable tasks.

• Leverage the system for complex workflows requiring high traceabilty, such as linking task completion to ADR compliance or automated stats generation.