← ClaudeAtlas

architecture-decision-advisorlisted

Power skill that forces a recorded, whole-repo architecture call on high-complexity requirements, with rejected alternatives named.
JeelVankhede/agentsmyth · ★ 0 · AI & Automation · score 72
Install: claude install-skill JeelVankhede/agentsmyth
# Architecture Decision Advisor ## Purpose For high-complexity requirements, force a big-picture architecture decision that considers the whole repo — not just the local task — and record the decision plus the alternatives that were rejected and why. Prevents a complex change from accumulating local, uncoordinated decisions that only make sense in isolation. This is a power skill, not a lifecycle phase. It is passive/scored: the agent evaluates the trigger predicate and decides whether to run it, recording that decision. ## Invocation Context Use this skill when the recorded `skill_scoring` trigger for `architecture-decision-advisor` evaluates true: `complexity_score >= 60 OR touches_contract OR new_surface`. Do not invoke it for requirements below that threshold with no contract or new-surface signal — a local, self-contained change does not need a whole-repo architecture call. ## What To Load **Foundation** (confirm in context; load if not already present): - Root `AGENTS.md` - `workflow/router.md` - `workflow/lifecycle.md` - `workflow/rules.md` **Minimum for invocation**: - This file - `references/output-schema.md` **Before starting work**: - `references/decision-framing.md` — how to frame alternatives so the record is genuinely useful later, not just a formality - `workflow/config/repo-profile.yaml` — repo mode, protected paths, public contracts - The requirement as currently framed, and `repo-alignment-scan`'s findings if it already ran this session ## Inputs