dundas/liveport

dundas/liveport

Agent Claude Code

Adversarial critique agent that validates implementation against requirements. Never implements; only validates, questions, and requires proof.

dundas/liveport

Agent Claude Code

Generate step-by-step configuration/installation/deployment tasks from an implementation plan, with prerequisites, deterministic commands, validation, and rollback.

dundas/liveport

Agent Claude Code

Break down high-level goals into minimal, verifiable sub-tasks with clear dependencies, checkpoints, and acceptance criteria.

dundas/liveport

Agent Claude Code

Validate features work end-to-end in production conditions. Never mark complete until the full user flow works.

dundas/liveport

Agent Claude Code

Implement safety-critical features with focus on edge cases, constraints, error handling, and defensive programming.

dundas/liveport

Agent Claude Code

Write precise specifications (OpenAPI, contracts, interface specs) from PRDs and plans with examples and constraints.

dundas/liveport

Agent Claude Code

Implement tasks with a strict red→green→refactor test-first workflow and small, reviewable commits.

dundas/liveport

Agent Claude Code

Turn a decomposed plan into a concrete technical approach, selecting patterns, modules, and interfaces that fit the repo.