Dispatching Parallel Agents

Dispatching Parallel Agents

Adapted from obra/superpowers — MIT. Adapted for the NodeJS-Starter-V1 monorepo (frontend/backend/database component separation).

Overview

When you have multiple unrelated failures across different components or files, investigating them sequentially wastes time. Each independent investigation can happen in parallel.

Core principle: Dispatch one agent per independent problem domain. Let them work concurrently. Integrate results after all complete.

When to Use

Use when:

• 3+ test files failing with different root causes
• Multiple independent subsystems broken (e.g., React component + FastAPI endpoint + DB model)
• Each problem can be understood without context from the others
• No shared state between investigations (agents won't edit the same files)

Don't use when:

• Failures are related (fixing one might fix others — investigate together first)
• Exploratory debugging (you don't know what's broken yet)
• Agents would edit the same files (shared state conflict)
• All failures clearly stem from a single root cause

---

The Pattern

Step 1: Identify Independent Domains

Group failures by component. In NodeJS-Starter-V1:

Each domain is independent — fixing a React component doesn't affect pytest failures.

Step 2: Create Focused Agent Prompts

Each agent gets:

• Specific scope: One test file, one component, or one subsystem
• Clear goal: Make these specific tests pass
• Constraints: Do NOT change code outside this scope
• Expected output: Summary of root cause and changes made

Good agent prompt template:

Fix the N failing tests in <test-file-path>:

1. "<test name>" — <error description>
2. "<test name>" — <error description>

Your task:
1. Read the test file and understand what each test verifies
2. Identify root cause — use systematic-debugging skill (Phase 1 first)
3. Write a failing regression test for each issue (TDD skill)
4. Fix the root cause, not the symptom
5. Verify all N tests pass: <test command>
6. Ensure no other tests regressed

Do NOT change code outside <scope-path>.
Return: Summary of root cause and what you changed.

Step 3: Dispatch in Parallel

In Claude Code, use the Task tool to dispatch concurrent agents:

Task("Fix apps/web/__tests__/components/auth-form.test.tsx failures — 3 failing tests [TDD + systematic-debugging]")
Task("Fix apps/backend/tests/api/test_contractors.py failures — 2 failing tests [TDD + systematic-debugging]")
Task("Fix apps/backend/tests/unit/test_models.py failures — 1 failing test [TDD + systematic-debugging]")
# All three run concurrently

Step 4: Review and Integrate

When all agents return: 1. Read each summary — understand what changed and why 2. Check for conflicts — did any agents edit the same files? 3. Run full test suite — verify all fixes work together

# Integration verification
pnpm turbo run test
pnpm turbo run type-check
pnpm turbo run lint

4. Spot check — agents can make systematic errors; review critical changes manually

---

Real-World Example

Scenario: CI fails after a backend refactor. 5 test failures across 3 files.

Failures:

• apps/web/__tests__/components/contractor-card.test.tsx: 2 failures (missing prop types after API change)
• apps/backend/tests/api/test_contractors.py: 2 failures (422 on new required field)
• apps/backend/tests/unit/test_validators.py: 1 failure (ABN validator edge case)

Decision: Independent domains — React component props separate from FastAPI validation separate from validator logic.

Dispatch:

Agent 1 → Fix contractor-card.test.tsx (frontend scope only)
Agent 2 → Fix test_contractors.py (backend API scope only)
Agent 3 → Fix test_validators.py (unit validator scope only)

Results:

• Agent 1: Updated TypeScript interface to include new abn field from API
• Agent 2: Added default value for new availability_status field in test data
• Agent 3: Fixed edge case: ABN with leading zeros was being trimmed

Integration: All fixes independent. No conflicts. Full suite green.

Time saved: 3 investigations in parallel vs. 3 sequential investigations.

---

Common Mistakes

---

Verification After Integration

After all agents complete and you've integrated:

# Must all pass before closing the task
pnpm turbo run test       # All tests green
pnpm turbo run type-check # 0 TypeScript errors
pnpm turbo run lint       # Lint clean

Use the verification-before-completion skill before claiming "Done" — run the commands, read the output.

---

Key Benefits

1. Parallelisation — Multiple investigations happen simultaneously 2. Focus — Each agent has narrow scope, less context noise 3. Independence — Agents don't interfere with each other 4. Speed — N problems solved in time of 1

---

Integration with Existing Infrastructure

• `/minion` command — Minion currently dispatches serially. This skill provides the formal pattern for concurrent dispatch when multiple independent failures are detected.
• `systematic-debugging` skill — Each agent should apply systematic-debugging within its scope (4-phase protocol, 3-attempt circuit breaker).
• `tdd` skill — Each agent writes failing tests first, then fixes.
• `verification-before-completion` skill — Final integration check before claiming all done.