Distill workflows from successful tasks or co-create custom Skills with QoderWork.
Beyond using ready-made Skills, you can create your own to capture personal expertise and best practices.
When QoderWork nails a task, ask it to save the workflow right in the conversation:
Have a clear workflow you want to automate? Start a co-creation session from the settings panel:
Each Skill is essentially a
Key field reference:
After creating a Skill, verify it with these steps:
Distill from a successful task
When QoderWork nails a task, ask it to save the workflow right in the conversation:
That turned out great — please save this workflow as a skill so I can reuse itQoderWork reviews the conversation, extracts the key steps, format requirements, and decision criteria, then generates a structured Skill file. After you confirm, it's saved and ready for the next similar task. No need to plan ahead — just capture what works when you see it.
Co-create with QoderWork
Have a clear workflow you want to automate? Start a co-creation session from the settings panel:
- In the left sidebar, open Extensions → Skills, then click Create with QoderWork.
- Describe your needs in plain language, e.g., "Every week I need to compile Slack project discussions into a report with three sections: progress, risks, and next week's plan."
- QoderWork generates a structured Skill file based on your description. Review, tweak if needed, and confirm — it's saved locally and ready to use.
Skill file format
Each Skill is essentially a SKILL.md file using YAML frontmatter + Markdown body format:
| Field | Required | Description |
|---|---|---|
name | Yes | Skill name — lowercase letters, digits, and hyphens |
description | Yes | Trigger description — QoderWork uses this to determine when to automatically invoke the skill |
version | No | Semantic version number |
Testing your Skill
After creating a Skill, verify it with these steps:
- Start a new conversation and describe a task that should trigger the Skill.
- Check automatic triggering — if it doesn't fire, refine the keywords in your
description. - Review the output format — make sure the output follows the expected steps and formatting.
- Try edge cases — describe the same type of task in different ways to confirm consistent triggering.

