Implement Feature

Maximum utilization of parallel subagent execution for feature implementation.

Quick Start

/implement user authentication
/implement real-time notifications
/implement dashboard analytics

Phase 1: Discovery & Planning

1a. Create Task List

Break into small, deliverable, testable tasks:

• Each task completable in one focused session
• Each task MUST include its tests
• Group by domain (frontend, backend, AI, shared)

1b. Research Current Best Practices

# PARALLEL - Web searches (launch all in ONE message)
WebSearch("React 19 best practices 2026")
WebSearch("FastAPI async patterns 2026")
WebSearch("TypeScript 5.x strict mode 2026")

1c. Context7 Documentation

# PARALLEL - Library docs (launch all in ONE message)
mcp__context7__query-docs(libraryId="/vercel/next.js", query="app router")
mcp__context7__query-docs(libraryId="/tiangolo/fastapi", query="dependencies")

Phase 2: Skills Auto-Loading (CC 2.1.6)

CC 2.1.6 auto-discovers skills – no manual loading needed!

Relevant skills activated automatically based on task:

• api-design-framework – REST/GraphQL patterns
• react-server-components-framework – RSC, Server Actions
• type-safety-validation – Zod, tRPC, Prisma
• unit-testing / integration-testing – Test patterns

Phase 3: Parallel Architecture Design (5 Agents)

Launch ALL 5 agents in ONE Task message with run_in_background: true:

# PARALLEL - All agents in ONE message
Task(subagent_type="Plan", prompt="...", run_in_background=True)
Task(subagent_type="backend-system-architect", prompt="...", run_in_background=True)
Task(subagent_type="frontend-ui-developer", prompt="...", run_in_background=True)
Task(subagent_type="llm-integrator", prompt="...", run_in_background=True)
Task(subagent_type="ux-researcher", prompt="...", run_in_background=True)

Phase 4: Parallel Implementation (8 Agents)

Phase 5: Integration & Validation (4 Agents)

Phase 5.5: Progress Notifications (CC 2.1.7)

CC 2.1.7 supports inline notification patterns for real-time progress updates:

Agent Completion Notifications

When subagents complete, track their progress:

# After parallel agent execution
for result in agent_results:
    notification.inline(f"{result.agent}: {result.status}")

# Summary notification
notification.inline(f"Phase 5 complete: {len(agent_results)}/{expected} agents finished")

MCP Deferral Awareness

When MCPs are deferred due to context limits, adapt your workflow:

if mcp.deferred:
    notification.inline("MCP tools deferred - using cached docs")
    # Fall back to cached documentation
    docs = load_cached_docs("react-19-patterns")
else:
    # Normal MCP query
    docs = mcp__context7__query_docs(...)

Progress Tracking Pattern

Use inline notifications for long-running phases:

[Phase 4] Starting: 8 parallel agents
[Phase 4] Complete: backend-system-architect (2.3s)
[Phase 4] Complete: frontend-ui-developer (3.1s)
[Phase 4] Complete: database-engineer (1.8s)
...
[Phase 4] Finished: 8/8 agents (12.5s total)

Phase 6: E2E Verification

If UI changes, verify with agent-browser:

agent-browser open http://localhost:5173
agent-browser wait --load networkidle
agent-browser snapshot -i
agent-browser screenshot /tmp/feature.png
agent-browser close

Phase 7: Documentation

Save implementation decisions to memory MCP for future reference:

mcp__mem0__add-memory(content="Implementation decisions...", userId="project-decisions")

Summary

Total Parallel Agents: 17 across 4 phases

Tools Used:

• context7 MCP (library documentation)
• mem0 MCP (decision persistence)
• agent-browser CLI (E2E verification)

Key Principles:

• Tests are NOT optional
• Parallel when independent (use run_in_background: true)
• CC 2.1.6 auto-loads skills from agent frontmatter
• Evidence-based completion

Related Skills

• explore: Explore codebase before implementing
• verify: Verify implementations work correctly

References

• Agent Phases