superlazy-plan

name	superlazy-plan
description	Silent planning and execution for multi-step tasks. Use when a task spans several files/steps, or when a written plan exists and must be executed.

Superlazy Plan

SUPERLAZY mode: output ≤3 lines (questions / user-actions / ✅❌); minimal code, full quality; do the work, never narrate it.

Writing plans

• Default: plan is INTERNAL — head or scratchpad file. Never shown to the user.
• Write a plan document ONLY when the work spans sessions/subagents or the user asks. Save to docs/plans/YYYY-MM-DD-<name>.md.
• Written plans: exact file paths, real code in every code step (no "TBD"/"add validation"/"similar to Task N"), exact commands with expected output, task = smallest independently testable unit, interfaces between tasks spelled out (exact signatures). DRY, YAGNI, TDD, frequent commits.
• Self-check before use: every spec requirement has a task; no placeholders; names/types consistent across tasks. Fix inline, don't re-review.

Executing plans

• Read the plan once, question it critically — contradictions or gaps → ONE batched blocking question before starting, not per-discovery interrupts.
• Execute every task in order without pausing to ask "continue?". Follow steps exactly; run every verification listed.
• Stop ONLY for: real blocker, genuinely ambiguous instruction, or all tasks done. Blocked → one-line question; done → ✅.
• Work spanning sessions → append per-task status to .superlazy/progress.md (same ledger as superlazy-subagents); trust ledger + git log over memory after compaction.
• Independent tasks + subagent support available → prefer superlazy-subagents.
• Never start on main/master without explicit user consent (see superlazy-worktree).