ulak.os

React/Next.js anti-pattern denetimi — state management, effects, re-render performansı, mimari sınırlar, client-side security ve accessibility eksikleri için yapılandırılmış audit. Mevcut bir React/Next.js yüzeyinde kod yazmadan ÖNCE veya /ulak-frontend war room'u sırasında "neyi yanlış yapıyoruz" sorusunu somut kontratlarla cevaplar. typescript-nextjs rule-pack'i frontend kalite-barında genişletir.

Read-only index of community Claude Code packs (commands / skills / agents / plugins / MCPs) curated from awesome-claude-code and related public sources. Produces a cross-referenceable pack-catalog with license + trust tier + capability-fit against Ulak OS gaps. Use during pack-gap audits, when onboarding a new plugin, or when answering "does someone else already solve this?". Does NOT install anything; read-only reference for operator decisions.

Auto-maintain the Ulak OS MCP allowlist. Reads `docs/governance/mcp-governance.md` + `.mcp.json` + `settings.local.json`, detects drift (newly added MCPs in settings that aren't governance-approved; approved MCPs not present; stale entries past rotation cadence), writes a reconciliation report, and proposes a PR-ready diff. Operator-gated (never auto-applies). Use during /ulak-director komple Phase 2 evidence pass, quarterly governance review, or when adding a new MCP needs a paper trail.

Materialize a full-stack SaaS project skeleton with Ulak OS governance pre-wired (Next.js + TypeScript + Supabase + optional payment + i18n + RLS + CI + tests + Traefik deploy). Generates a shippable starter that embeds 23 sector packs + 8 rule packs + 79 anti-patterns from commit 1. Use when starting a new SaaS product OR when /ulak-scaffold command dispatches.

Read-only watchlist for external skill/plugin ecosystems that may inform Ulak OS capabilities. Use when evaluating new skill/plugin candidates from the inbox list, planning marketplace compatibility, or checking whether an external pack should be studied before building native Ulak OS functionality. Does not install third-party code.

Run a 14-dimension audit scorecard on any repo (Architecture, Testing, Secrets, Observability, CI/CD, Duplication, Dependencies, Type Safety, Plugins, API Design, Infrastructure, Frontend, Data Validation, Documentation). Produces per-dimension 0-100 scores, target state, gap analysis, and A-F grade. Use for project baselines, post-modernization verification, or quarterly health reports.

Extract and decompose a "god module" (single file >1000 LOC with 5+ unrelated responsibilities) using the Strangler Fig protocol. Produces new modules in target package, a thin re-export shim of the original, per-step atomic commits, and updated imports across the codebase. Use for monolithic files blocking modernization.

Orchestrate N specialist agents across a multi-sprint backlog using the merge-sequence protocol (infrastructure → backend → frontend → tests). Produces sprint assignments, dependency graph, file-ownership matrix, merge sequence, daily standup template. Use when planning a Waves-pattern execution involving 4+ agents.

final validation workflow for release-readiness, residual risk classification, and single-verdict closure after the main program work is done.

detect and explain missing reusable units such as commands, skills, agents, hooks, MCP connectors, docs, or evals when the project needs stronger automation and reuse.

intake, inventory, and evidence-gathering workflow for full audits, rescue programs, or when the user wants the system to read the whole project before acting.

research and currency check workflow for market signals, competitor positioning, language opportunities, and whether the proposed stack or UX is below the 2026 bar.