Human-AI Collaborative Coding Workflow

You are a workflow engineer. Build a human-AI collaborative coding workflow that interleaves autonomous agent work with human review and judgment. Tech Stack: [INSERT - FastAPI/NestJS + React UI + Slack/IDE integrations] LLM Agents: [INSERT - planner/coder/reviewer agents through the internal gateway] Orchestration: [INSERT - Temporal/LangGraph durable workflows] Channels: [INSERT - VS Code extension, Slack bot, web UI] Deliver the following modules in order: 1. Architecture: a durable workflow with explicit states for autonomous work, human decision gates and handoffs; a shared task and context store. 2. Data Model: task, proposal, decision and comment records; editable agent context; roles (driver/reviewer/approver) and SLA timers. 3. Interaction APIs: request task, review diff, approve/request-changes/edit-and-resume, notifications with deep links into IDE or Slack. 4. Collaboration UX: streaming agent progress, inline diff review, comment threads on code, suggested edits, autonomy-level settings per task risk. 5. Handoff & Context: agent-generated handoff brief (what it did, open questions, how to verify); full context preserved when a human takes over driving. 6. Testing: workflow state-machine tests, UI end-to-end tests with mocked agents, timeout/escalation tests, acceptance-rate and rework metrics.