Course handbook

SDD with SPECTRA

Spec-driven development for engineers who already write code daily and have at least passing exposure to AI coding agents. You learn the method by running it — the core loop, a greenfield and a brownfield build, and the workflows that sit around it: deciding whether an idea is worth building, fixing what breaks, and shaping the process to fit your team. The labs run on SPECTRA; the method transfers to any Spec Kit project.

Total time About 13 hours
Pace Self-paced or 3-day workshop
Format 10 modules + 5 labs
SPECTRA and Spec Kit.SPECTRA is built on top of GitHub Spec Kit — it does not replace it. Spec Kit provides everything this course teaches as method: the spec-driven loop (constitution, specify, clarify, plan, checklist, tasks, analyze, implement, converge), the bug-fix and idea-assessment workflows, and the extensions, presets and overrides you customize it with. SPECTRA installs and keeps that toolchain current with one command, and adds its own agents for the work around the code — requirements, impact, architecture decisions, test strategy, root-cause analysis, and pull-request delivery. Everything except the speckit.spectra.* agents transfers to a plain Spec Kit project. SPECTRA is built and maintained by TELUS Digital — see the SPECTRA landing page for the full picture.
Command spelling.Every command has one identity — written speckit.specify or speckit.spectra.adr — and your coding agent decides how you type it. The course shows the form Claude Code and GitHub Copilot use: /speckit-specify, /speckit-spectra-adr. Other agents differ: kiro-cli uses /speckit.specify, Codex uses $speckit-specify, and Kimi uses /skill:speckit-specify. Your agent’s own command list always shows the exact trigger.

Course structure

1
SDD Foundations
What a spec is, the full spec-driven workflow and its human gates, how specs persist and evolve, when not to use SDD
Reading + knowledge check
50 min
2
Anatomy of a Good Spec
The six-element framework, what belongs in the spec versus the plan, checklists as unit tests for requirements, constitution vs spec
Reading + writing exercise
60 min
3
Greenfield Lab — SPECTRA from Scratch
Install SPECTRA and run the full workflow — with clarify, checklist, analyze, and converge — on a new to-do app
Hands-on lab
2 hrs
4
Brownfield Lab — Add “Due Date” to the To-Do App★
The centerpiece. Adopt the workflow in an existing codebase: a reviewable baseline, guardrails drawn from the code, one bounded change.
Hands-on lab
2–3 hrs
5
Idea Assessment — Should We Build It?
Take an idea from intake to a go, needs-clarification, or kill verdict backed by evidence — before anyone writes a spec
Concepts + hands-on lab
75 min
6
Bug Fix — Assess, Fix, Verify
Diagnose a real defect in the lab app, fix only what the diagnosis covers, and verify against the original reproduction
Concepts + hands-on lab
75 min
7
Customization — Make the Workflow Your Own
Extensions, presets, project overrides, workflows, and bundles — and SPECTRA’s own template overrides and artifact root
Concepts + hands-on lab
60 min
8
Best Practices & Pitfalls
What works consistently, sizing work to the context window, what sabotages adoption, three deliberately broken specs to critique
Reading + critique exercise
60 min
9
Quality, Testing & Team Collaboration
Why specs catch what unit tests miss, the role split, the review-bottleneck risk
Reading + team discussion
60 min
10
AI-DLC
Inception → Construction → Operation, how the agentic SDLC phases map onto them, bolts instead of sprints, and who is in the room at each gate
Reading + knowledge check
60 min

★ Module 4 is the centerpiece. If you only do one lab, do this one.

Suggested learning paths

Self-paced (recommended)
One or two modules per week over about six weeks. Gives the labs time to settle and the discussion prompts time to land in real team contexts.
Three-day workshop
Day 1: Modules 1–3 (foundations and the greenfield lab). Day 2: Modules 4–5 (the brownfield lab and idea assessment). Day 3: Modules 6–10 (bug fix, customization, practice, team, and AI-DLC).
Engineering-leadership track
Modules 1, 5, 8, 9, and 10 — read the concepts and skip the labs. Covers everything needed to sponsor adoption: the method, how ideas earn investment, what goes wrong, and how the lifecycle itself changes.
Hands-on track
The five labs — Modules 3, 4, 5, 6, and 7. For engineers who want to feel the practice before reading about it.

Lab codebase

Every lab uses the same kind of app — Next.js 15 + TypeScript + Tailwind CSS + localStorage, no backend — so what you learn in one lab transfers directly to the next. (The app you scaffold in Module 3 gets whatever create-next-app installs today, React 19 and Tailwind CSS v4; the Modules 4–7 starter pins React 18 and Tailwind CSS 3.)

The starter ships as lab-codebase.zip; the download link is on each lab page. Unzip it and run npm install && npm run dev to get started.

Prerequisites

Use the course navigation to begin with Module 1 — SDD Foundations.