Pipeline Mode

Pipeline mode runs multi-stage coding workflows: Plan → Explore → Code → Review, with parallel fan-out. Assign different models to different stages — Opus for planning, Qwen for coding.

Setup

Pipeline mode requires Python 3.11+:

cd pipeline && pip install -e .

Usage

cloding pipeline "Add authentication" --workspace ./myapp --no-docker

With a specific config:

cloding pipeline -c configs/qwen-fanout.yaml "Refactor the DB layer"

Pipeline Stages

A typical pipeline runs through these stages:

1. Plan — Analyze the task and create a development plan
2. Explore — Examine the codebase to understand the current state
3. Code — Implement the changes (can fan out to multiple parallel agents)
4. Review — Review the changes for correctness and quality

Each stage can use a different model. Use a powerful model for planning and a fast, cheap model for execution.

Built-in Configs

8 pipeline configurations are included:

Config	Description
default	Standard 4-stage pipeline
quick	Fast 2-stage: plan + code
fan-out	Parallel coding with multiple agents
opus-plan+qwen-code	Opus for planning, Qwen for coding
human-in-the-loop	Pauses for human approval between stages
And more...	Check pipeline/configs/ for all options

Custom Pipelines

Create a YAML config to define your own pipeline:

stages:
  - name: plan
    model: opus
    prompt: "Analyze and plan: {task}"
  - name: code
    model: qwen
    parallel: 3
    prompt: "Implement: {plan}"
  - name: review
    model: sonnet
    prompt: "Review these changes: {code}"

Place your config in the pipeline/configs/ directory and reference it with -c.