Test agents

You are helping a developer validate agent triggering and functionality through systematic testing.

Test scope: $ARGUMENTS

Workflow

1. Load skill context - Run oaps skill context agent-development --plugin --references triggering system-prompts to get detailed guidance

2. Determine scope - Based on arguments:

   • If agent name provided: test that specific agent
   • If empty: list agents and offer to test specific ones

3. Read agent definition - Read the agent file to understand:

   • Frontmatter configuration
   • Triggering examples
   • System prompt structure
   • Tool restrictions

4. Validate structure - Run oaps agent validate <name> to check configuration:

   • Report any parsing errors
   • Note any warnings
   • Fix structure issues before proceeding with functional tests

5. Analyze triggering examples - Review the description for:

   • What explicit requests should trigger the agent
   • What proactive scenarios should trigger it
   • What variations should work

6. Test explicit triggering - For each example in the description:

   • Note the expected user message
   • Note the expected assistant response
   • Document how to test manually

7. Test phrasing variations - Plan tests with variations:

   • Rephrase requests differently
   • Use synonyms for key terms
   • Try shorter/longer versions

8. Test proactive triggering - If agent should trigger proactively:

   • Identify prerequisite actions
   • Plan statements that should trigger without explicit request

9. Test negative cases - Verify agent does NOT trigger when:

   • Request is similar but out of scope
   • A different agent should handle it

10. Test functionality - Once triggered, verify:

    • Agent follows its defined process
    • Meets quality standards
    • Produces correctly formatted output
    • Handles edge cases

11. Present results - Summarize test outcomes:

    • Triggering tests (explicit, variations, proactive, negative)
    • Functional tests
    • Edge cases that need attention
    • Recommendations for improvements

12. Document findings - If issues found, offer to:

    • Launch agent-developer agent to add triggering examples or clarify system prompt
    • Create todo items for tracking
    • Suggest using /oaps:agent:write for rewrites