Skip to content

architecture-review

Internal forked reviewer sub-agent (context: fork, model: opus) for the strategy workflow. Acts as a platform architect assessing refined strategy features in artifacts/strat-tasks/ (cross-referenced against source RFEs) for architectural correctness: verifies every dependency against the architecture docs, checks integration patterns match how components actually communicate, ensures component boundaries aren't violated, validates the deployment model, and flags architectural conflicts and cross-component coordination needs (versioning, rollout order, backwards compatibility). Applies architecture-context overlays as authoritative corrections, skips cleanly when no architecture context is available, and grounds every finding in specific docs, components, or APIs.

Plugin: rfe-creator | Internal

Contract

Skill Contract canonical-skill-v1

Review strategy features for architectural correctness — dependencies, integration patterns, and component interactions.

Identity

Functions
review
Success
  • Produces an architecture assessment for each strategy with a recommendation.
  • Grounds every finding in architecture docs with specific component and API citations.

Optimization Targets

task_success judge SKILL.md @ c8a2a1e

Invariants

Must Preserve
  • Do not approve strategies with incorrect dependency assumptions.
  • Do not flag concerns without citing specific architecture documentation.
Fixed Context
toolsRead, Grep, Glob
knowledgerepository_contentpublic, task_inputtask_private

Traceability

Skill
.claude/skills/architecture-review/SKILL.md

Diagram

architecture-review diagram

Usage

/architecture-review