Overview
The same task can be completed in different ways. Qoder CLI CN offers multiple ways of working, covering scenarios from instant conversation to long-running automation. Picking the right one lets you strike the right balance between "controlling every step" and "letting Qoder finish autonomously."
This page helps you choose quickly based on the nature of the task. Each way of working has its own detailed documentation you can click into.
Quick picker
- Want to work as you chat and intervene anytime → Interactive mode
- Want to see the plan clearly and confirm before changing anything → Plan
- Have a clear goal and want Qoder to keep going until it is done → Goal
- Need to run at a specified time, or repeat on a calendar → Scheduled tasks
- Need to run the same thing repeatedly at fixed intervals → /loop
- Need multiple Agents collaborating on a complex flow → Workflow
- Need automated runs in scripts or CI/CD → Headless
Work as you chat: Interactive mode
Interactive mode is the most common default. Run qodercn in the project root to enter an interactive (TUI) session, converse in natural language, and watch Qoder's actions, confirm permissions, and correct its direction at any time.
Best for tasks that need close human–AI collaboration, such as exploring code, implementing features step by step, and debugging. See Running Tasks.
Plan before executing: Plan
Use Plan when the task is complex or far-reaching, or when you want to see the approach clearly before touching anything. In Plan, Qoder explores the code read-only and produces a plan without modifying code directly; once you confirm the plan, exit Plan and start executing.
Enter or exit via the /plan command. See Plan Before Executing.
Work continuously toward a goal: Goal
Use Goal when you have a clear completion condition and want Qoder to keep working autonomously until it is met. Enter via /goal <goal>; Qoder automatically switches into an automated execution state to reduce interruptions along the way.
Great for tasks with a clear finish line, like "get all tests green" or "finish this refactor until the build passes." See Working Toward a Goal.
Run at a specified time: Scheduled tasks
Use scheduled tasks when a task needs to run once at some future time, or repeat on a calendar schedule. Good for periodic inspections, recurring reports, and automation triggered on a schedule.
See Scheduled Tasks.
Repeat at fixed intervals: /loop
Use /loop when you need to run the same task repeatedly at fixed time intervals. Good for polling status, continuous checks, and periodic repeated actions.
See Repeating Tasks.
Multi-agent collaboration: Workflow
When a task is complex enough to need multiple Agents dividing the work, you can orchestrate a workflow that splits the task across different Subagents. Good for decomposing large tasks and driving them forward in parallel.
For creating and orchestrating Subagents, see Built-in Capabilities; for multi-agent orchestration, see Dynamic Workflows; for parallel collaboration, see Parallel Tasks.
Run in scripts: Headless
Use Headless (non-interactive) mode when you need to embed Qoder CLI CN into scripts, automated processes, or CI/CD pipelines. Run via qodercn --print to output structured results as Text, JSON, or Stream JSON.
Since nobody is around to confirm, permissions need to be configured in advance in Headless mode. See Run in Scripts.
Combining them
These ways of working are not mutually exclusive — they can be combined. For example:
- Confirm the approach with Plan first, then enter Goal and let Qoder execute autonomously to completion.
- Pair Headless mode with a permission mode for unattended yet controlled automation.
- Use scheduled tasks or /loop to trigger a fixed automation routine.