One developer with Claude Code is a workflow. Five is a coordination problem. Connect Jira, Confluence and Figma over MCP, write tickets agents can execute, and keep review, metrics and incidents under control.
A team readiness assessment with an adoption level and action items
A working .mcp.json for Jira, Confluence or Figma, or a filesystem MCP stand-in
An agent-friendly ADR template and one real ADR
A reusable spec-driven ticket template and one feature built from a ticket
A review pipeline in your repo: pre-review checks, shared .githooks/, review checklist
Understand the adoption spectrum, new roles that emerge, how SDD principles scale to teams, domain ownership, and critical anti-patterns to avoid.
Install and configure MCP servers for Jira, Confluence, and Figma so Claude Code can read tickets, query docs, and inspect designs directly.
Structure Confluence as a living spec store with agent-friendly ADRs, project context optimization, and traceability back to SDD artifacts.
Write Jira tickets as executable specs with acceptance criteria, linked ADRs, and automated pipelines from ticket to PR to deploy.
Convert Figma designs to code with Claude Code — extract design tokens, map components, generate accessible markup, and maintain design-code traceability.
Manage 3-5x PR volume with branch strategies, automated pre-review, adversarial review checklists, and review SLAs that prevent bottlenecks.
Coordinate across timezones with handoff docs and session summaries, automate standups, track velocity and quality metrics, and build the ROI case.
Handle agent hallucinations, wrong merges, spec drift, and production incidents from AI-generated code — the playbook for when things go wrong.
Follow a team through a full sprint: Figma design to specs, specs to Jira tickets, Claude Code implementation, review, deploy — the complete SDD pipeline.
One year of access for $20, or subscribe for $45/month to get this course and every other one.
Course 1 (Claude Code Fundamentals) and Part 1 of Course 3 (SDD foundations). This course builds on the spec-first habits and ADRs from Course 3, applied at team scale.
No. They are used in the examples, but each hands-on has a fallback. Without access, you simulate the same pattern with the filesystem MCP server and local markdown files. Some exercises assume a team of 2+ developers.
9 tutorials, each with exercises and a knowledge check. The listed lesson times add up to about 6 to 8 hours. The readiness hands-on is a 30 to 45 minute team session.
This course is about people: roles, tickets, review, metrics and incidents when a team works with agents. Software Factories builds an unattended issue-to-PR pipeline. In both, a human reviews and merges every PR.