Resillion

Internal cheat sheet

AI Prompting Cheat Sheet

A practical reference for writing clearer prompts, improving output quality, and refining responses quickly. Designed for Resillion teams working across delivery, assurance, and Total Quality engagements.

Be specific Add context Set constraints Define format Give examples Iterate

Core Prompt Formula

Role + Task + Context + Constraints + Output Format

You are a QA consultant.
          Review the following user story for ambiguity, missing acceptance criteria, and testability gaps.
          Keep the response concise and present the output in a table with columns:
          Issue, Why it matters, Recommendation.

Prompt Structure

A reliable prompt usually answers five questions: who, what, why, limits, and output shape.

Role
Who should the AI act as?

Task
What exactly should it do?

Context
What background or situation matters?

Constraints
What should it avoid or stay within?

Output Format
How should the response be structured?

The 6 Things That Improve Prompt Quality

Principle	Weak	Better
Be specific	Write a test plan.	Create a high-level test plan for a SaaS login and registration feature, covering scope, risks, environments, test types, and entry/exit criteria.
Add context	Explain this.	This is for a senior stakeholder review, so keep the language professional and non-technical.
Set constraints	Tell me everything.	Keep it under 300 words, use UK English, and focus only on functional risks.
Define output format	Give me the answer.	Return the response in a 3-column table: Issue, Impact, Recommendation.
Give examples	Use a good structure.	Use this format for each recommendation: Recommendation / Benefit / Effort.
Iterate	Accept first response.	Refine with follow-ups like: make this more concise, rewrite for executives, add examples.

Reusable Prompt Template

Use this when you want consistent, reusable results.

Help me [objective].
          Context: [background].
          Audience: [who this is for].
          Constraints: [limits or rules].
          Format: [table/email/bullets/etc.].
          Tone: [tone].

Tip: good prompts are usually clearer when written like a mini brief rather than a loose request.

Prompt Patterns That Work Well

Summarise Rewrite Analyse

Summarise
Summarise the following defect trend report into 5 key insights for a delivery lead.

Rewrite
Rewrite this test exit report so it is clearer, more concise, and suitable for senior stakeholders.

Analyse
Review this requirement and identify ambiguity, missing acceptance criteria, and testability gaps.

Show more prompt patterns

Compare Generate Transform Critique Classify Extract Plan Prioritise

Compare
Compare manual and automated regression options in terms of effort, speed, coverage, and maintenance.

Generate
Generate 10 test scenarios for this checkout flow, including positive, negative, and edge cases.

Transform
Turn these workshop notes into a structured test plan with scope, risks, environments, and milestones.

Critique
Challenge this proposed automation strategy and identify weaknesses, assumptions, and missing controls.

Extract
Extract all acceptance criteria, business rules, and integrations from the following user story.

Plan
Create a phased test approach for this release, including entry criteria, dependencies, and reporting checkpoints.

Prioritise
Prioritise these defects based on business risk, user impact, and release readiness.

Prompting Frameworks

A.P.E

Action + Purpose + Expectation

Review this test strategy, so I can send feedback to the client, and provide the
              output as a professional email response.

R.A.C.E

Role + Action + Context + Expectation

You are a Test Architect. Assess this QA operating model. The organisation has mixed
              agile maturity and inconsistent governance. Provide a prioritised list of improvements.

C.A.R.E

Context + Action + Result + Example

We are assessing QA maturity across multiple teams. Review these findings and
              produce a summary of the top risks, expected business impact, and example recommendations.

C.O.S.T.A.R

Context + Objective + Style + Tone + Audience + Response

Context: We are preparing for a release readiness review for a payments platform.
              Objective: Create a concise outcome report. Style: Structured and evidence-based. Tone: Professional and
              direct. Audience: Senior stakeholders and delivery leads. Response: Return a summary with overall status,
              key risks, evidence gaps, and recommended next steps.

Good vs Bad Prompt

Bad

Write something about testing.

Too vague. No audience, purpose, scope, or structure.

Good

You are a QA lead. Write a one-page overview of why test data management matters in large organisations with multiple systems. Include key challenges, common risks, and 5 practical recommendations. Use clear business language.

Clear role, topic, scope, expected depth, and style.

Refine the Output

Follow-up prompts are often where the best results come from.

Make this more concise Rewrite in plain English Turn this into a client-ready email Add examples Challenge your own answer Convert this into a slide summary

Common Mistakes

Being too vague
The AI cannot guess the right angle if the request is broad.

No context
Without background, the model fills in gaps with assumptions.

No output format
Unstructured answers are harder to review and reuse.

Accepting the first response
Strong prompting is usually iterative, not one-shot.

Power Prompt Formula

Act as [role].
          Help me [task].
          Here is the context: [context].
          Keep in mind: [constraints].
          Return the answer as [format].
          The audience is [audience].
          Use a [tone] tone.

Example Prompts for Work

Use case	Prompt
Generate tests	Act as a QA analyst. Generate functional, negative, and edge-case test scenarios for this feature based on the provided requirements and acceptance criteria.
Create a test plan	Act as a Test Lead. Create a high-level test plan for this release, including scope, assumptions, risks, test types, environments, dependencies, and entry/exit criteria.
Outcome reporting	Act as a QA lead. Create a concise test outcome report summarising execution progress, defects, outstanding risks, coverage achieved, and release readiness.
Requirement review	Review the following requirement and identify ambiguity, missing acceptance criteria, testability concerns, and any questions that should be resolved before delivery begins.
Defect triage support	Analyse these defects and group them by likely root cause, business impact, and urgency, then recommend which should be fixed before release.
Automation planning	Act as an SDET. Recommend which of these regression scenarios should be automated first based on frequency, risk, stability, and expected return on investment.