t-planlisted

# T-Plan Skill You are the orchestrator here, not the implementer. Exploration, alternatives research, and documentation validation each belong to a subagent, so the plan is assembled from independent findings rather than from one context that already decided the answer. Start by creating the session directory (`mkdir -p .t-plan/<session-id>`), writing `intent.md` to capture the request, and creating the master task with `TaskCreate`. The artifacts are the point — the user invoked `/t-plan` for the full orchestrated workflow, not ad-hoc help. --- Transform conversations into rock-solid implementation plans, coordinating subagents with the `Agent` tool and tracking their work with the `Task*` tools. ## Architecture Overview ``` No hooks required - Agent dispatch plus the Task* tools handle all coordination natively. Orchestrator responsibilities: - Session initialization (mkdir, .gitignore, current.txt pointer) - Task lifecycle (TaskCreate, TaskUpdate) - Pre-truncation before subagent dispatch - Output verification after subagent return - Retry logic (max 2 attempts) Subagent responsibilities: - Read context files (intent.md, explore.md, etc.) - Write output files directly (explore.md, scout.md, validation-vNNN.json) - Focus on their assigned task ``` **Reference material** (load when needed): - Retry logic & parallel execution: See `references/subagent-patterns.md` - Planning principles & anti-patterns: See `references/planning-principles.md` - Plan output template: