Saga

When you need Claude to autonomously build a medium-to-large feature end-to-end, this spins up an orchestrator-worker architecture that treats specs like contracts. It forces you to nail down every validation criterion up front in a hierarchical spec tree (saga, milestones, tasks), then delegates implementation to worker subagents while keeping the orchestrator's context clean. The bet is simple: if the spec is airtight enough that a worker can self-verify against it, you can parallelize work and step away. It writes everything to a persistent directory under ~/.sagas so you can resume across sessions. Best for features that would normally fragment your context across dozens of turns.

skills.sh installs: 20.8K
GitHub stars: 684
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add warpdotdev/common-skills --skill saga --agent claude-code

Installs into .claude/skills of the current project.

Files

Saga

Saga is an autonomous, spec-driven development workflow for medium-to-large features that should be implemented mostly without human intervention, except at a few discrete touch points. You act as the orchestrator: you turn a rough prompt into an airtight spec, then delegate implementation to a fleet of worker subagents while keeping your own context window clean.

The whole method rests on one bet: if the spec defines every task with validation criteria tight enough to form a contract, then workers can execute in parallel and self-verify, and the saga succeeds with almost no human babysitting. The quality of the saga is therefore decided in Phase 1, before a single line is written.