ccw-chain
CCW Chain orchestrates multi-step workflows by analyzing user intent, routing to the appropriate chain, and executing skill pipelines through progressive step loading. It supports auto mode for hands-off execution, delegation between chains, and variable propagation across the entire workflow.
CCW Chain routes user intent to workflow chains and executes skill pipeline steps progressively.
AI-generated summary based on this skill's SKILL.md
Install
catlog22/Claude-Code-Workflow/ccw-chain · repository language: TypeScript
git clone https://github.com/catlog22/Claude-Code-Workflow
cp -r Claude-Code-Workflow/.claude/skills/ccw-chain ~/.claude/skills/ccw-chainnpx skillfed install catlog22/Claude-Code-Workflow/ccw-chainFrequently asked questions
AI-generated answers based on this skill's SKILL.md and metadata
How do I orchestrate a workflow chain with ccw-chain?
CCW Chain orchestrates multi-step workflows by analyzing your intent and routing you to the appropriate chain. Once routed, it executes skill pipelines through progressive step loading, meaning each phase loads and completes before the next begins. You can enable auto mode to bypass confirmation prompts and let the chain run hands-off, or manually step through each phase. Variables and context propagate across the entire workflow so downstream steps have access to upstream results.
What is ccw-chain workflow routing and how does it work?
CCW Chain workflow routing analyzes your intent and directs you to the correct workflow chain and entry point. The skill examines what you're trying to accomplish, matches it against available chains, and selects the best starting node. This routing ensures you enter the right pipeline for your task. Once routed, the chain executes its steps sequentially, with each phase completing before the next begins, and context flowing through the entire workflow.
Can ccw-chain automate skill pipeline execution with auto mode?
Yes. CCW Chain supports auto mode, which automates skill pipeline execution by bypassing confirmation prompts at each step. When auto mode is enabled, the chain runs hands-off from start to finish without waiting for your approval between phases. Context and variables propagate automatically across all steps, allowing downstream skills to use upstream results. This is ideal for complex multi-step workflows where you want uninterrupted execution.
How do I discover available chains and select entry points?
CCW Chain provides list and inspect commands to discover available chains and preview their node graphs. Use the list command to see all chains, then inspect a specific chain to view its structure, phases, and available entry points. This lets you understand the workflow before execution and choose where to start. You can also preview how variables flow through the chain before committing to execution.
What is chain delegation and how does ccw-chain handle it?
CCW Chain supports delegation between chains, allowing one workflow to hand off execution to another. When a chain delegates, it passes its current context and variables to the target chain, which then continues the work. This enables complex multi-phase skill pipelines where different chains specialize in different tasks. Delegation depth and variable propagation are managed automatically, so context flows seamlessly across chain boundaries.
How does ccw-chain execute phase content inline within skill chains?
CCW Chain executes phase content inline by loading and running each phase's steps directly within the skill-level chain. As each phase completes, it produces artifacts—results or outputs—that are handed off to the next phase. This progressive execution model ensures phases run in sequence with full context awareness. Inline execution keeps everything within the chain's scope, avoiding unnecessary context switching between separate tools or systems.
SKILL.md
rendered from the published skill — quoted content, verbatim
CCW Chain Orchestrator
Chain-based workflow orchestrator using chain_loader for progressive step loading and LLM-driven decision routing.
Discovery
chain_loader list— list all chains with triggers, entries, and descriptions- Match user intent to chain
triggers.task_types/triggers.keywords chain_loader inspect— preview chain node graph and available entrieschain_loader start— begin from default entry, named entry (entry_name), or any node (node)
Execution Protocol
When chain_loader delivers a step node with a skill/command doc:
- Read the loaded doc content to understand the skill's purpose and interface
- Assemble the Skill call:
Skill(skill_name, args) - First step:
args = "${analysis.goal}" - Subsequent steps:
args = ""(auto-receives session context) - Special args noted in step name
(truncated - see the full file via the links below)
Read as markdown · JSON record · Browse the source repository
File tree — 10 files
.claude/skills/ccw-chain/SKILL.md
.claude/skills/ccw-chain/chains/ccw-cycle.json
.claude/skills/ccw-chain/chains/ccw-exploration.json
.claude/skills/ccw-chain/chains/ccw-issue.json
.claude/skills/ccw-chain/chains/ccw-lightweight.json
.claude/skills/ccw-chain/chains/ccw-main.json
.claude/skills/ccw-chain/chains/ccw-standard.json
.claude/skills/ccw-chain/chains/ccw-team.json
.claude/skills/ccw-chain/chains/ccw-with-file.json
.claude/skills/ccw-chain/specs/auto-mode.md