← Back to LLM prompts

Convert Bug Reports into Agile User Stories with Gherkin Acceptance Criteria, Scaled by Complexity

Transforms raw bug reports into sprint-ready user stories using the 'As a/I want to/So that' format and Given/When/Then acceptance criteria, automatically adjusting detail level based on bug complexity (severity, reproducibility, impact). Ideal for backlog refinement and sprint planning.

data a general-purpose LLM WritingBusiness
<role>
Senior Agile Business Analyst and QA Specialist with expertise in backlog refinement, behavior-driven development (BDD), and adaptive story slicing.
</role>

<instructions>
Convert each bug report in [bug_report_input] into a well-formed Agile User Story with Gherkin-style Acceptance Criteria. Follow these steps:

1. Analyze the bug report for: role impacted, desired behavior, business value, severity, reproducibility, data dependencies, and edge cases.
2. Determine complexity level: Low (cosmetic, single-step, isolated), Medium (functional deviation, multi-step, config-dependent), High (data corruption, security, race conditions, cross-system).
3. Write the User Story in format: "As a [specific role], I want to [clear action], so that [measurable benefit]."
4. Write Acceptance Criteria using Given/When/Then structure. For Low complexity: 1–2 scenarios. Medium: 3–5 scenarios covering happy path + 1–2 edge cases. High: 6+ scenarios including boundary values, error states, permissions, rollback, and data integrity checks.
5. Tag each story with Complexity: [Low|Medium|High].
6. Optionally suggest Story Points based on [team_velocity_reference] if provided.
7. Ensure all output is in English, free of markdown, and formatted for direct import into Jira/Azure DevOps.
</instructions>

<context>
Project: [project_context]
Team Velocity Reference (optional): [team_velocity_reference]
Bug Reports to Process:
[bug_report_input]

Each bug report should contain: Title, Description, Steps to Reproduce, Expected Result, Actual Result, Severity, Environment, Affected Component.
</context>

Process the provided bug reports now and output the formatted user stories.
Website Source
#text