Writing Effective Tasks for AI Agents

Your task title should describe what needs to be done in action-oriented language. Start with a verb like "Add", "Create", "Fix", or "Update".

Good examples:

• "Add user authentication to the login page"
• "Fix the broken image upload feature"
• "Create a dashboard for viewing statistics"

Avoid vague titles like:

• "Work on login" (too vague)
• "Bug" (doesn't describe the problem)

The description should explain why this task matters and what the expected outcome is. Include:

• The problem you're solving
• Who benefits from this change
• Any relevant context or background
• Links to related resources or discussions

The more context you provide, the better the AI agent can understand your intent.

Acceptance criteria tell the AI agent exactly what "done" looks like. Use simple, testable statements:

• "Users can log in with email and password"
• "Error messages display when login fails"
• "Session persists after browser refresh"

Think of acceptance criteria as a checklist that can be verified.

Help the AI agent understand how to prioritize work:

Priority levels:

• Critical: Blocking issues, security problems
• High: Important features, significant bugs
• Medium: Normal feature work
• Low: Nice-to-haves, minor improvements

Complexity:

• Small: Simple changes, under 2 hours
• Medium: Moderate effort, 2-8 hours
• Large: Significant work, 8+ hours