{"name":"mikeparcewski-wicked-garden","owner":{"name":"ClaudePluginHub"},"plugins":[{"name":"mikeparcewski-wicked-garden","source":{"source":"github","repo":"mikeparcewski/wicked-garden"},"description":"AI-Native SDLC — executable infrastructure, not passive guidance. Requires wicked-testing (npm) as a peer plugin for QE behavior. Leverages Claude Code's native surface: TaskCreate metadata envelope validated by PreToolUse, 75 specialists routed dynamically by subagent_type frontmatter, skills with progressive disclosure, 14 lifecycle hook events. A facilitator skill scores 9 factors, detects 1 of 7 project archetypes, and picks specialists + phases per project. Hard quality gates with BLEND multi-reviewer aggregation, convergence lifecycle tracking, challenge gate + contrarian at complexity ≥ 4, phase-boundary QE evaluator with per-archetype evidence demands, semantic reviewer for spec-to-code alignment, and 3-tier persistent memory with auto-consolidation. 13 domains across the full lifecycle: crew workflows, context assembly, memory, code intelligence, brainstorming, and domain experts for security, architecture, QE, product, data, delivery, agentic, and personas. Runs with local SQLite storage; wicked-bus (npm) recommended for event-driven gate verdicts.","version":"7.1.0","strict":true,"keywords":["sdlc","workflow","engineering","testing","delivery","memory","search","brainstorming","personas","code-intelligence","agentic"],"category":"productivity"}]}