do:deepen-plan

Plan enhancement is automated by orchestrating parallel research, skill, and review agents. All agent outputs are persisted to disk, keeping the parent agent's context lean and preventing context exhaustion during long sessions.

Use automated plan enhancement to dynamically discover available research and review agents. These agents execute in batches to analyze architecture and identify vulnerabilities, writing comprehensive findings directly to disk for later review.

Yes, persistent outputs are retained across runs for traceability and learning. Each execution gets its own numbered directory on disk, ensuring all research findings and plan improvements are saved without filling up memory.

Agent orchestration manages execution in batches with robust error handling and stats capture. It dynamically discovers available skills and learnings, coordinating a multi-phase synthesis and triage process for comprehensive plan improvement.

No, automated discovery dynamically finds available skills, learnings, and review or research agents. You simply point the skill to your existing plan file, and it handles the parallel agent execution and multi-phase synthesis automatically.