Give Claude a concrete task, relevant source material and a clear definition of a useful result. Review the answer and refine it using specific feedback.
This is a practical prompting guide, not a ranking or a claim that one tool suits every task. Anthropic’s prompting documentation starts with defining success criteria and evaluating results; your own task provides the useful test.
A practical workflow.
Define success
Write a short description of the result you want. For a summary, that could mean accurate decisions and unanswered questions rather than a polished but vague paragraph.
Supply relevant material
Provide only information you are permitted to share. Make the boundaries between your instructions and the source text clear.
Specify the output
Ask for a structure that supports your task: a short brief, a comparison table or a list of questions. Include an example if it clarifies the format.
Evaluate and revise
Compare the answer with your success criteria and the original source. Ask for a specific correction and verify the revised result.
READY TO MAKE YOURS
Start with this prompt.
Using only this source material: [text], create [output] for [audience]. A successful result must [criteria]. Separate established facts from unanswered questions. Return [format], preserve [important details] and flag unsupported assumptions.
Replace the brackets. Leave out private details. Check the response.
ILLUSTRATIVE EXAMPLE
A clearer request.
A VAGUE STARTSummarise this document perfectly.
A CLEARER BRIEFSummarise this document for a colleague who needs the decisions and unresolved questions. Use only the supplied text and label anything the document does not establish.
The brief defines the audience and protects the source boundaries.
Before you use the result.
- The output matches your success criteria.
- Claims are supported by the supplied material.
- Unknowns remain unknown.
- Sharing complies with your rules.
A few useful answers.
Can I use the same prompt in different AI tools?
A clear text instruction is a useful starting point across tools. Test the result and adjust for the actual behaviour and features of the tool you use.
How much context should I include?
Start with the material that affects the task. More text is not automatically more useful if it obscures the goal or includes unrelated information.
Original sources & further reading.
Check current product guidance when a feature or setting matters to your task.
