Prompt Engineer Susanne Renate Schneider: AI Skills Should Be Manually Triggered First, Automate After Verifying Timing
Prompt engineer Susanne Renate Schneider advises on the t3n MeisterPrompter podcast that ChatGPT and Claude Skills should initially be configured to trigger only manually, observing which scenarios warrant automation before gradually opening them up. She also shares a starter prompt leveraging memory functions to identify tasks suitable for Skill creation, and warns that OpenAI will discontinue CustomGPTs service on December 11.
Skills in ChatGPT and Claude are not more complex prompts; they are reusable work instructions. A Skill specifies the concrete steps for a certain type of task, the expected output, and the context in which it should be invoked. Creating one requires no programming skills; natural language descriptions suffice. Prompt engineer Susanne Renate Schneider believes the standard for judging whether a Skill is worth the effort is straightforward: Are you repeatedly performing the same type of task, always relying on the same criteria for judgment?
While Skills can be set to trigger automatically, Schneider recommends starting with manual triggers. The reason is: only after you have personally invoked them a few times can you see which scenarios were meant for the Skill to intervene automatically and which merely occur by chance. She suggests configuring Skills to require explicit commands to start initially, observing for a period, and then gradually enabling automation, adjusting the trigger conditions at that point.
For users unsure where to begin, Schneider shares a starter prompt she has used herself: 'Of the tasks I complete here in [insert specific AI tool], which are worth creating a Skill for?' This prompt relies on the AI tool's memory function being enabled; its role is to filter through your existing conversation history to identify recurring task patterns, helping you find the first batch of scenarios worth solidifying into Skills.