This project demonstrates an end-to-end AI-assisted development workflow using spec-kit with the Spec Kit plugin for GitHub Copilot CLI. Starting from a raw GitHub issue, an idea was assessed, specified, planned, and shipped entirely through structured agents — first the Idea Assessment Pipeline (the assess extension), then Spec-Driven Development. The steps below document exactly what was entered — you can follow the same process to take your own ideas from issue to implementation.
Note on the workflow: The feature was not driven by an elaborate formal specification. It began as a raw GitHub issue, was assessed through the intake → research → define → shape → decide pipeline, and only then handed off to Spec-Driven Development. Each command below is the exact prompt entered into Copilot Chat, as you will see.
This project is built on top of spec-kit, an open-source toolkit for Spec-Driven Development created and maintained by GitHub. All credit for the toolkit, its agents, and its extensions belongs to the spec-kit contributors. Please visit the original repository to learn more, contribute, or show your appreciation.
-
The Spec Kit
specifyCLI on yourPATH:uv tool install specify-cli # or: pipx install specify-cli specify --version -
A GitHub repository to work in
This demo is driven by the Spec Kit Copilot plugin, which exposes the specify CLI to the Copilot agent as skills. Register the marketplace, then install the plugin:
copilot plugin marketplace add github/spec-kit-copilot
copilot plugin install spec-kit-copilot@spec-kit-marketplaceVerify it loaded:
copilot plugin list
# or, inside an interactive session:
/skills listSee the plugin README for local development installs and troubleshooting.
The assess extension adds the Idea Assessment Pipeline: five skills (intake, research, define, shape, decide) that turn a raw idea into a scored decision before any code is written.
Initialize the project as a Spec Kit project using Copilot and skills mode
This scaffolds .specify/ and installs the spec-kit commands as Copilot Agent Skills (.github/skills/speckit-*/SKILL.md).
/skills reload
Loads the newly scaffolded skills into the running session.
Add the assess extension
Installs the Idea Assessment Pipeline — the five assess skills (intake, research, define, shape, decide).
/skills reload
Picks up the newly added assess skills.
Now do an intake using https://github.com/github/spec-kit/issues/3423
Captures the raw idea from the issue into .specify/assessments/specs-of-specs/intake.md.
Continue
Gathers evidence for and against the idea into research.md.
Continue
Frames the problem — affected users, goals, non-goals, success metrics — in problem.md.
Continue
Sketches 2–3 solution options with appetite and trade-offs in concept.md.
What about D a separate page describing option 4 in more detail?
Adds Option D (a dedicated docs page) and makes it the recommended option.
Continue
Scores the idea and records the verdict in decision.md: go, handing off to Spec-Driven Development.
With a go decision in hand, the standard spec-kit agents took the idea from specification to a shipped change.
Continue
Converts the go verdict into specs/001-spec-of-specs-docs/spec.md plus a spec quality checklist.
Yes
Produces plan.md and design artifacts (research.md, data-model.md, contracts/, quickstart.md).
Yes
Generates the dependency-ordered tasks.md (14 tasks, grouped by user story).
Do the implementation
Executes the tasks: adds docs/concepts/spec-of-specs.md, trims Option 4 in complex-features.md to a summary plus link, and wires the new page into docs/toc.yml.
A raw GitHub issue turned into a shipped change — assessed, specified, planned, and implemented through structured agent workflows without manual scaffolding. In this run the idea happened to resolve to a documentation change (a new concept page wired into the docs table of contents), but the point of the demo is the workflow itself: the Spec Kit plugin for GitHub Copilot CLI carrying any idea from issue to implementation.