← ClaudeAtlas

decompose-project-lotslisted

Split an approved mission into small, ordered, independently verifiable lots with dependencies, exclusions, proof, and stop conditions. Use when asked to break down, phase, sequence, or audit a project that is too large for one safe execution pass.
kotnisofiane-bit/dubsar-agent-skills · ★ 0 · AI & Automation · score 66
Install: claude install-skill kotnisofiane-bit/dubsar-agent-skills
# Decompose Project Lots ## Objective Create the smallest ordered sequence that still produces meaningful, human-reviewable progress. ## Inputs - an approved `mission.json`; - known dependencies, protected areas, and irreversible boundaries; - existing evidence identifiers, if any. ## Lot design rules Each lot must: - have one primary outcome; - name the files, systems, or artifacts it may touch when known; - list dependencies and required prior evidence; - define what it will not do; - specify a proportional validation method; - end at a human-reviewable boundary; - avoid bundling cleanup, feature work, deployment, and migration together. ## Workflow 1. Read the approved `mission.json`. 2. Identify dependency order and irreversible boundaries. 3. Prefer discovery or preservation lots before mutation lots. 4. Give each lot a stable ID and one sentence outcome. 5. Record expected evidence and stop conditions. 6. Mark only the first unblocked lot as `candidate`; leave all others `planned`. ## Output Write `lots.json`: ```json { "format": "dubsar.project-lots/1", "mission_id": "same-as-mission", "lots": [ { "lot_id": "local-stable-id", "title": "bounded outcome", "depends_on": [], "in_scope": [], "excluded": [], "expected_evidence": [], "validation": [], "stop_conditions": [], "status": "planned|candidate|complete" } ] } ``` ## Limits - Do not alter the approved mission or execute any lot. - Keep