What happens if a run is interrupted?

The skill persists state to a session directory. You can resume the process exactly where it left off using the --from flag followed by the stage name.

What is the CWF run skill?

It is an orchestration tool that chains multiple specialized CWF skills into a single, end-to-end automated pipeline for software development tasks.

Does it work on my current git branch?

By default, if you start a run on your base branch (like main), it automatically creates and switches to a new feature branch to protect your codebase.

How does it handle ambiguous requirements?

It uses configurable Ambiguity Modes. 'Strict' mode stops and asks the user for every decision, while 'defer-blocking' allows the AI to make autonomous choices that must be reviewed before merging.

Can I skip specific stages in the workflow?

Yes, the skill supports a --skip flag allowing you to bypass specific stages like 'ship', 'retro', or 'review-plan' to customize the pipeline execution.

CWF Workflow Orchestrator

Name: CWF Workflow Orchestrator
Author: corca-ai

bycorca-ai

•

生産性とワークフロー

Orchestrates the complete software development lifecycle by auto-chaining requirement gathering, planning, implementation, and shipping into a single autonomous pipeline.

The CWF Run skill is the master orchestration layer for the Corca Workflow framework, designed to handle end-to-end task delegation. It eliminates the need for manual stage sequencing by automatically driving a project through a standardized nine-stage pipeline: from initial requirement gathering and planning to implementation, automated code review, and final shipping. Built with context-deficit resilience, it uses persistent session states to ensure the workflow can recover from interruptions or be resumed from specific stages, all while maintaining strict human-in-the-loop gates for critical pre-implementation decisions.

主な機能

01Automated quality loops including review, refactor, and retrospective stages

02State-persistent execution allowing resumption from any stage via --from

03Configurable ambiguity modes (Strict, Defer, Explore) for autonomous decision-making

0473 GitHub stars

05Full SDLC auto-chaining from gathering requirements to shipping PRs

06Deterministic pre-flight readiness checks to ensure environment stability

ユースケース

01Standardizing development workflows across teams using a deterministic stage pipeline

02Resuming long-running coding tasks from a specific failure point or review stage

03Executing a complex feature request from a single prompt without manual intervention

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add corca-ai/claude-plugins run

For use in Claude.ai and ChatGPT

Download Skill