slopstop

Ticket-anchored AI development for Linear, JIRA, and GitHub Issues, built on one idea: stop slop before it goes in, instead of reviewing it out afterwards. Work starts from a ticket, not a prompt. Claude writes failing tests for what the ticket requires — not for what the code already does — then implements against a written scope boundary and asks before wandering outside it. Nothing merges without a simplify pass over the uncommitted diff and an adversarial review that checks every finding against the real code, and each ticket keeps a durable plan, findings, and progress log outside the repo so a fresh session resumes where the last one stopped. Preventing slop does not mean working alone. The same guarantees scale to a fleet: :design interviews you into a PRD, :tickets cuts an adversary-approved ticket tree from it, and :run drives parallel headless agents — one per ticket, each isolated in its own git worktree — toward that tree, across four model tiers where every tier's work is checked by the tier above it, with frozen-test tamper checks, independent handoff verification before any branch is integrated, and a human gate at each stage boundary. Eighteen commands in all — more details at https://github.com/iansmith/slopstop

19 skills · 5 agents