Shiplight Quality

Try Quality on a project (in Claude Code)

From the repository you want to assess, install the quality agent skill for Claude Code:

cd /path/to/your/project
npx skills add ShiplightAI/quality/agent-skills \
  --skill quality --agent claude-code -y

The command above installs the skill only for Claude Code. To install it for several agents, list them after --agent (for example, --agent claude-code cursor); to install it for every supported agent, use --all instead.

Then open Claude Code in that repository and give it this prompt:

/quality start

The agent will inventory the repository, propose a small quality graph for the highest-priority feature, connect the proof that already exists, and run an initial assessment when runtime results are available. It will pause when a human needs to confirm feature boundaries, priorities, or quality checks.

After the first pass, inspect the proposed files under .quality/ and answer the agent's review questions. Then continue with the next command it recommends, usually /quality improve or /quality map-feature <feature>.

Want to control the first slice? Include it in the prompt:

/quality start, focus on the highest-risk user-facing workflow first

The skill evaluates existing specifications, tests, workflows, reports, and other evidence; it does not create tests or approve its own proposals. See the quality skill for all commands and workflow details.

What Quality does

Quality connects product intent to independent, repeatable proof:

requirements (PRDs)
        ↓
features (specs)
        ↓
quality checks
        ↓
proof definitions (tests and other evidence)
        ↓
runtime observations (did the proof pass?)

Requirements define desired outcomes, checks state what must hold, and tests, workflows, telemetry, or manual records provide proof. Quality maps these relationships in .quality/ and keeps missing or weak links visible. Agents can propose maps; people review structure and accepted risk.

The engine reports four separate measures:

Measure	Question it answers
Quality	Is the current proving evidence passing?
Coverage	Does every declared check have proof?
Evidence confidence	Is the mapped proof strong enough?
Structure confidence	Are these the right features, checks, and priorities?

See the concepts guide for the model and trust boundaries.

Optional: Use spec-project for spec-driven testing

spec-project is an optional companion for turning product intent into accepted feature specs and aligned testing evidence. It produces artifacts; quality independently evaluates them. The skills can be used together or separately.

Install the optional spec-project skill for Claude Code:

npx skills add ShiplightAI/quality/agent-skills \
  --skill spec-project --agent claude-code -y

Then invoke /spec-project init, /spec-project lifecycle, or /spec-project maintenance. The installer creates agent-specific files and skills-lock.json in the target project.

For the quality graph model and independence guarantees, see the concepts guide and how Quality earns your trust.

Development

Requirements: Node.js 24 or newer and pnpm 11.

Install dependencies and run the repository checks:

pnpm install
pnpm test
pnpm typecheck
pnpm build

To inspect another local repository with Quality Explorer, start the application with an absolute project path. For GitHub Actions observations, authenticate with gh auth login first and pass the token; omit the first line for local-only observations:

GITHUB_TOKEN="$(gh auth token)" \
QUALITY_PROJECT_ROOT=/absolute/path/to/project \
  pnpm --filter @shiplightai/quality-explorer dev

The token is passed only to the local process. dev starts the Next.js development server with hot reload; it does not run an assessment or deploy the explorer.

Then open http://127.0.0.1:4173/quality-explorer. The project root is fixed when the process starts and cannot be changed by a browser request.

License

Shiplight Quality is available under the MIT License. See LICENSE.