Claude Code Skills — Full Delivery Lifecycle

This is not a single skill — it is a complete software delivery methodology packaged as 129 Claude Code skills across 7 plugins, with 3 bundled MCP servers. Created by Lev Nikolaevich, the suite automates the full lifecycle from project bootstrap through documentation, Agile planning, task execution, quality gates, performance optimization, and community management. The architecture follows an Orchestrator-Worker hierarchy with four levels: L0 meta-orchestrator drives the full pipeline, L1 coordinators manage major lifecycle stages, L2 coordinators handle scope within stages, and L3 workers execute specific tasks. Each level has single responsibility and loads only the context it needs, keeping token usage efficient even on large projects. The 7 plugins cover distinct lifecycle phases: agile-workflow (scope decomposition, story management, execution, quality gates), documentation-pipeline (auto-detect project type, generate full docs), codebase-audit-suite (security, build, code quality, tests, architecture, performance — 35+ audit skills), project-bootstrap (create or transform projects to Clean Architecture), optimization-suite (performance profiling, dependency upgrades, modernization), community-engagement (GitHub triage, announcements, RFC management), and setup-environment (CLI agent installation, MCP config, settings sync). Three MCP servers extend agent capabilities beyond what skills alone can do. hex-line-mcp provides hash-verified file editing with 11 tools — every edit includes a content hash to prevent stale-context corruption. hex-graph-mcp indexes codebases into a SQLite graph via tree-sitter AST with 15 tools for finding references, cycles, clones, unused exports, and architecture overviews. hex-ssh-mcp enables token-efficient remote SSH editing with hash verification across 6 tools. The workflow moves through a clear pipeline: ln-010 sets up the development environment, ln-100 generates all documentation, ln-200 decomposes scope into Epics and Stories, ln-1000 orchestrates execution through tasks, validation, execution, and quality gates. Human approval checkpoints are built in at validation and quality gate stages. Multi-model AI review is a standout feature. Code and story reviews can be delegated to Codex and Gemini agents running in parallel, with automatic fallback to Claude Opus. This catches issues that a single model might miss and reduces review bottleneck time. Related: For DNS infrastructure management skills, see the easyDNS MCP server. For document processing in your pipeline, check Chandra OCR.