Creating Tasks for Jules
The quality of Jules' output depends heavily on how well you describe your task. Learn to write clear, actionable task descriptions that lead to high-quality pull requests.
Anatomy of a Good Task
A well-written task description gives Jules everything it needs to produce the right solution. Include these elements:
-
Clear Objective
State exactly what you want done. Be specific about the expected outcome, not just the problem.
-
Context
Explain why this change is needed, what the current behavior is, and what the desired behavior should be.
-
Scope
Define which files or components should be changed, and which should not be touched.
-
Acceptance Criteria
List specific conditions that must be true for the task to be considered complete.
Good vs Bad Task Examples
"Fix the login bug."
This is too vague. Which login bug? What is the current behavior? What should happen instead? Which files are involved?
"Fix the login form validation bug where users can submit the form with an empty email field. The email input in src/components/LoginForm.tsx should show a validation error message when the field is empty and the user clicks Submit. Add a unit test in tests/LoginForm.test.tsx to verify this behavior."
Task Templates
Use these templates as starting points for common task types:
## Bug Description [What is broken and when does it occur?] ## Current Behavior [What happens now?] ## Expected Behavior [What should happen instead?] ## Steps to Reproduce 1. [Step 1] 2. [Step 2] 3. [Step 3] ## Files Involved - src/components/MyComponent.tsx - src/utils/helpers.ts ## Acceptance Criteria - [ ] Bug is fixed - [ ] Existing tests still pass - [ ] New test added for the fix
## Feature Description [What new functionality should be added?] ## Requirements - [Requirement 1] - [Requirement 2] - [Requirement 3] ## Implementation Notes [Any specific patterns, libraries, or approaches to use] ## Files to Create/Modify - Create: src/features/newFeature.ts - Modify: src/index.ts (add export) ## Acceptance Criteria - [ ] Feature works as described - [ ] Unit tests written and passing - [ ] Follows existing code patterns
Providing Context
Jules reads your codebase, but additional context helps it make better decisions:
- Link related issues: Reference other GitHub issues that provide background
- Mention coding patterns: "Follow the same pattern used in UserService.ts"
- Specify libraries: "Use the
date-fnslibrary for date formatting, notmoment" - Note constraints: "Must be backward-compatible with the existing API"
- Include error messages: Copy exact error text from logs or stack traces
Monitoring Progress
Once you assign a task, Jules provides real-time updates on its progress:
| Status | Meaning |
|---|---|
| Queued | Task is waiting to be picked up by Jules |
| Analyzing | Jules is reading the codebase and planning the approach |
| Implementing | Jules is writing code changes |
| Testing | Jules is running tests to validate the changes |
| PR Created | A pull request is ready for your review |
| Failed | Jules encountered an issue and could not complete the task |
Reviewing Pull Requests
When Jules creates a pull request, treat it like any other PR from a team member:
- Read the PR description: Jules explains what it changed and why
- Review the diff: Check the code changes for correctness and style
- Run the tests: Ensure all tests pass in your CI pipeline
- Test manually: Verify the changes work as expected in your environment
- Request changes: If something needs adjustment, comment on the PR and Jules may be able to iterate
💡 Try It: Write a Task Description
Pick a real bug or feature request from one of your projects. Write a task description using the templates above. Include a clear objective, context, scope, and acceptance criteria. Then assign it to Jules and observe the results.
Ready to Go Deeper?
Live instructor-led courses from our partners. Affiliate disclosure.
AI & ML Courses - 30% Off
Live instructor-led AI, machine learning, data science, and cloud courses for working professionals. Use code Limited30 at checkout.
EdurekaDataCamp - AI & Data Science
Hands-on Python, machine learning, and AI courses with interactive exercises and real projects.
DataCampedX - Top AI Courses
University-level AI courses from MIT, Harvard, Stanford. Earn certificates that employers recognize.
edX