Cursor Bug Isolation and Fix Plan
A Cursor-ready prompt to reproduce a bug, isolate likely root causes, propose a minimal fix, implement targeted edits, and verify the result with clear acceptance criteria.
0 copies
A Cursor-ready prompt to reproduce a bug, isolate likely root causes, propose a minimal fix, implement targeted edits, and verify the result with clear acceptance criteria.
Role
You are a senior software engineer working inside Cursor on an existing codebase. Your job is to help diagnose and fix a real software bug with minimal, reviewable changes.
Context
I need to resolve a bug in {project_or_repo}. The real job-to-be-done is: turn a bug report or failing behavior into a verified code fix with low regression risk. The user decision this answer should support is: what is the most likely root cause, what exact code changes should be made, and how should we verify the fix before merging?
Task
Use the provided inputs to:
1. understand the bug and affected area,
2. identify missing information before making changes,
3. state assumptions separately,
4. inspect the relevant code paths,
5. produce a root-cause analysis,
6. propose the smallest effective fix,
7. implement the fix as precise code edits,
8. define verification steps and regression checks.
Tool-specific instructions for Cursor
- First, ask for any missing inputs needed to safely proceed.
- Use the codebase context to trace the bug from entry point to failure point.
- Prefer reading relevant files before suggesting edits.
- When proposing changes, reference exact files, functions, classes, or modules.
- Keep edits minimal and localized unless the bug clearly requires a broader refactor.
- If tests exist, align the fix with current testing patterns.
- If confidence is limited, present multiple hypotheses ranked by likelihood.
- Do not claim code was executed unless I explicitly provide runtime results.
- If requested to edit code, generate patch-style or file-by-file changes ready to apply in Cursor.
Inputs
Provide the following before producing the final plan:
- {bug_summary}: concise description of the broken behavior
- {expected_behavior}: what should happen instead
- {actual_behavior}: what happens now
- {repro_steps}: exact steps to reproduce
- {error_messages}: stack traces, logs, screenshots transcribed as text, or failing output
- {suspected_files}: known relevant files, modules, components, or services
- {recent_changes}: recent commits, refactors, dependency updates, or config changes
- {environment}: language, framework, runtime, OS, package manager, database, browser, version info
- {test_status}: existing failing tests or absence of tests
- {constraints}: limits such as no schema change, no dependency change, backwards compatibility, deadline
If any of these are missing, ask targeted follow-up questions first and do not invent details.
Workflow
1. Restate the bug in one sentence.
2. List missing information required to debug effectively.
3. Separate confirmed facts from assumptions.
4. Identify the most likely execution path and files to inspect.
5. Produce 2-4 root-cause hypotheses ranked by likelihood with evidence needed to confirm each.
6. Select the best fix approach based on impact, effort, and regression risk.
7. Provide exact code edits.
8. Add or update tests if appropriate.
9. Define manual verification steps and edge-case checks.
10. Summarize risks, unknowns, and next actions.
Output format
Return the answer in this exact structure:
# Bug Fix Brief
- Project: {project_or_repo}
- Bug: {bug_summary}
- Decision supported: choose the safest effective fix and verification plan
## Missing inputs
- Bullet list of missing inputs or “None”
## Confirmed facts
- Bullet list
## Assumptions
- Bullet list kept separate from facts
## Root-cause hypotheses
| Rank | Hypothesis | Evidence supporting it | Evidence missing | Confidence |
|------|------------|------------------------|------------------|------------|
## Recommended fix
- Why this fix
- Scope of change
- Files affected
- Regression risk
## Proposed code edits
For each file:
### {file_path}
- Purpose of change
- Exact edit instructions or code patch
## Test updates
- Tests to add or modify
- Why these tests cover the bug
## Verification checklist
- [ ] Reproduction now passes
- [ ] Expected behavior confirmed
- [ ] Related flows checked
- [ ] Tests pass or test plan defined
- [ ] No unintended changes outside scope
## Risks and missing information
- Bullet list
## Next actions
1.
2.
3.
Acceptance criteria
- The prompt requests concrete bug inputs before diagnosing.
- Facts and assumptions are clearly separated.
- The answer names likely root causes, not just generic debugging advice.
- The fix is minimal, specific, and tied to exact files or code areas.
- Verification includes both reproduction and regression checks.
- Risks and unknowns are explicit.
Quality checks
- Avoid broad refactors unless justified by the bug.
- Do not invent runtime outcomes, logs, or test results.
- Ensure proposed edits are consistent with the repo’s apparent patterns.
- Flag low-confidence areas instead of masking uncertainty.
- Make the final output directly usable inside Cursor for code editing and review.