← ClaudeAtlas

architect-proposelisted

Manual-only skill. Use only when the user explicitly invokes the architect-propose skill to turn one already approved design bundle into one new independent sealed plan package under `.architect/`. One approved design bundle may contain multiple D-xxx subdesigns, and one resulting plan may contain multiple T-xxx tasks. Do not auto-trigger from a design or implementation request.
vortezwohl/Architect · ★ 27 · AI & Automation · score 72
Install: claude install-skill vortezwohl/Architect
# Architect Propose Use this skill only as the manually selected `architect-propose` stage. Its job is to store an already approved design bundle as a structured execution package and initialize the package state and execution log. It does not redesign, implement code, or invoke sibling skills automatically. It is stage 2 of the one-way flow `architect-design -> architect-propose -> architect-build`. ## Defined Terms - `approved design bundle`: the complete approved output from one `architect-design` invocation. - `D-xxx subdesign`: one approved architectural decision inside that bundle. - `independent plan`: one new future `.architect/<plan-name>/` package created from exactly one approved design bundle. - `T-xxx task`: one recorded execution task inside that independent plan. A single independent plan may contain multiple `T-xxx` tasks derived from one or more approved `D-xxx` subdesigns. - `functional boundary completion`: the packaging pass that turns approved design intent into target functionality, protected related functionality, explicit non-goals, compatibility obligations, and a hard-stop condition. - `code impact scope`: the expected paths, symbols, configuration, tests, and callers likely to be affected. It is a broad execution reference, not a hard source modification limit. - `functional boundary gap`: any missing target functionality, protected related functionality, explicit non-goal, compatibility obligation, or hard-stop condition that