Claude Code plugin — structural guardrails for multi-agent software development. RTM-based evidence, cross-model adversarial audit, 7 MCP tools, worktree isolation, HITL retrospective. DOI: 10.5281/zenodo.19108370
Agent
Headless worker for quorum — receives task + context, implements code, runs tests, submits evidence to watch file, handles audit corrections. Use when the orchestrator needs to delegate a coding task to a worker agent.
Agent
Read-only RTM generator — reads all track work-breakdowns, verifies each requirement against the actual codebase using deterministic tools, and produces 3 Requirements Traceability Matrices (Forward, Backward, Bidirectional). Use when the orchestrator needs to establish or update the RTM before distributing work.
Agent
Find UI issues that code-level analysis (scout, codemap, dependencygraph) cannot detect. Launches a real browser to check rendering, visual states, interactions, a11y, and runtime errors. Use after FE implementation to catch issues invisible to static analysis.