Idea Assessment Quickstart

Use this process to decide whether an idea deserves investment before committing to a solution. It works for software ideas and non-software decisions alike. No existing source code is required.

The bundled, opt-in assess extension provides five stages: intake → research → define → shape → decide. The result is an evidence-backed go, needs-clarification, or kill decision, not an implementation.

Set up

Install Spec Kit and initialize a project. An empty project is enough, or you can use an existing repository when the idea relates to its code. In a terminal at the initialized project root, install the extension:

specify extension add assess

Then launch your coding agent in that directory. The examples below use GitHub Copilot's default skills mode (--integration copilot). Other agents use their own command invocation syntax. Invoke each /speckit-assess-* skill separately in your agent's chat, reviewing the artifact before moving on. These are agent skills, not terminal commands; the terminal command above only installs the extension.

1. Capture the idea

/speckit-assess-intake "Let users work offline and sync when they reconnect." slug=offline-mode

You can also supply a URL, ticket, or codebase pointer. The agent captures the input in .specify/assessments/offline-mode/intake.md. The slug is the shared handle for all five stages.

2. Research the evidence

/speckit-assess-research slug=offline-mode

The agent gathers supporting and opposing evidence in research.md. Review sources and confidence levels: uncited assumptions are not established facts. This stage should challenge the idea, not just build a case for it.

3. Define the problem

/speckit-assess-define slug=offline-mode

The agent writes problem.md, describing the affected users, problem, goals, non-goals, success metrics, and cost of doing nothing.

4. Shape possible solutions

/speckit-assess-shape slug=offline-mode

The agent compares concept-level options, constraints, appetite, and trade-offs in concept.md. Detailed architecture, implementation tasks, and source changes are deliberately outside this process.

5. Make the decision

/speckit-assess-decide slug=offline-mode

The agent records the scorecard, rationale, and verdict in decision.md.

Verdict What it means What to do
go The problem and evidence are adequate and a concept is recommended Decide whether to act; optionally hand a software idea to SDD
needs-clarification Specific unknowns still block a defensible decision Refine the affected artifacts with the missing information
kill The idea is not worth pursuing now Keep the recorded rationale; stopping is a successful outcome

All five artifacts live under .specify/assessments/offline-mode/. None of the assessment commands modify source code.

Resolve unknowns by refining the artifacts

The usual flow runs each command once. When an artifact contains [NEEDS CLARIFICATION: ...], edit that Markdown file directly or ask the agent in free-form chat to incorporate the missing information:

Update .specify/assessments/offline-mode/problem.md with these confirmed success metrics: <your metrics>. Update any dependent wording and revise decision.md if this resolves the blocker. Preserve source and confidence tags.

Do not automatically regenerate the entire stage. Ask the agent to review whether the new facts clear the blocker and revise the downstream scorecard, verdict, or handoff accordingly. Rerunning a command is an exception, such as replacing a discarded draft, and existing artifacts must not be overwritten without confirmation.

Optional handoff to SDD

Assessment is standalone: it does not automatically invoke SDD or insert itself into the specification process. If you choose to build a software idea with a go verdict, establish your project constitution if needed, then pass the decision's handoff summary to the specification command:

/speckit-specify Use the handoff summary in .specify/assessments/offline-mode/decision.md to specify the approved offline-mode concept.

Continue with the SDD quickstart. For stage prerequisites, slug rules, and evidence requirements, see the idea assessment command reference.