Orchestrates multi-agent parallel test execution across isolated workspace branches using custom subagent definition and concurrent dispatch.
Workflow Steps
1. Register Custom Worker Subagent (define_subagent)
Invoke define_subagent to register a specialized worker subagent persona:
- name:
test_suite_runner - description: "Specialized test runner and diagnostic subagent that executes test suites, parses stack traces, isolates root causes, and generates fix recommendations."
- system_prompt: "You are a specialized Test Runner and Diagnostic Subagent. Execute assigned test suites in your workspace branch, parse failure stack traces, inspect affected source files to isolate root causes, and write a structured execution report to
reports/tests/<suite_name>.md. Return a Markdown summary with pass/fail counts, key stack traces, and actionable root-cause diagnoses." - enable_write_tools:
true - enable_mcp_tools:
false - enable_subagent_tools:
false
2. Dispatch Parallel Workers (invoke_subagent)
Identify target test suites from the user request or project configuration (e.g., Unit, Integration, E2E, API Contract, Security).
Dispatch all workers concurrently in a single invoke_subagent call with "Workspace": "branch":
- TypeName:
"test_suite_runner" - Role:
<Suite Name> Runner - Prompt: Specify the test command to execute, stack trace parsing directives, root-cause inspection rules, and report path (
reports/tests/<suite_name>.md). - Workspace:
"branch"
3. Aggregate & Report Test Results
Upon receiving completion callbacks from all background subagents:
- Verify execution reports exist under
reports/tests/. - Parse overall pass/fail metrics, execution durations, and root-cause failure stack traces.
- Present a consolidated test suite matrix to the user with actionable diagnostic summaries and clickable report links.