MCAF is a framework for building software products together with AI coding agents.
Skill Claude CodeCodex
Write or update an architecture decision record with alternatives, consequences, approval, and an executable implementation contract. Use for material design decisions or an explicit ADR request.
Skill Claude CodeCodex
Improve backlog readiness, delivery roles, team cadence, and feedback agreements around a concrete process problem. Use for team delivery policy, not an individual feature implementation plan.
Skill Claude CodeCodex
Create or refresh docs/Architecture.md as a concise map of real modules, boundaries, contracts, slices, and entry points. Use for onboarding or architecture documentation changes.
Skill Claude CodeCodex
Create or revise CI/CD workflows, required quality gates, versioned artifacts, release promotion, and rollback. Use for pipeline or delivery automation changes.
Skill Claude CodeCodex
Review a diff for correctness, requirement coverage, test quality, and operational risk, or refine PR review policy and templates. Use for concrete findings and review readiness.
Skill Claude CodeCodex
Fix local setup, run, debug, test, or onboarding friction using reproducible repository commands. Use when contributor workflows are missing, inconsistent, or slow.
Skill Claude CodeCodex
Update durable engineering docs, navigation, and canonical source placement when they are stale, duplicated, or missing. Use for general docs work; feature specs and ADRs have dedicated skills.
Skill Claude CodeCodex
Create or update an executable feature specification with stable requirements, acceptance criteria, real flows, ADR decisions, and test traceability. Use before non-trivial behavior changes or on an explicit spec request.
Skill Claude CodeCodex
Prioritize human review of a large generated code area by tracing real flows and ranking risky files. Use when a reviewer cannot inspect every line and needs a practical review sequence.
Skill Claude CodeCodex
Plan or improve data exploration, ML feasibility, experiments, evaluation, inference, and model operations. Use for training, inference, or data-science delivery, not generic AI coding.
Skill Claude CodeCodex
Turn relevant quality goals into measurable requirements and verification for performance, reliability, accessibility, data integrity, or other system attributes.
Skill Claude CodeCodex
Instrument or specify logs, metrics, traces, correlation, and actionable alerts for a concrete runtime flow. Use when detection or diagnosis is missing or a change affects operational visibility.
Skill Claude CodeCodex
Review or implement scoped security controls for authentication, authorization, secrets, external inputs, data handling, and pipeline permissions. Use for a concrete security impact or baseline request.
Skill Claude CodeCodex
Refactor coupled or oversized code by responsibility, apply repository maintainability limits, and document justified exceptions. Use for design quality changes or maintainability policy.
Skill Claude CodeCodex
Create or update root and project-local AGENTS.md, repository boundaries, rule precedence, skill routing, and agent workflow policy. Use for MCAF bootstrap or governance changes.
Skill Claude CodeCodex
Define or refine branch, merge, commit, release, and repository ownership policy. Use for source-control governance or secrets-in-history response, not routine Git commands.
Skill Claude CodeCodex
Design or update regression tests for features, defects, and refactors using repository verification commands. Use for meaningful user or system flow coverage and test strategy.
Skill Claude CodeCodex
Define or refine UI engineering guidance, design-system handoff, accessibility, and frontend technology decisions. Use for concrete UI delivery or design-policy changes.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: