No. The meditation app is the example threading the site together — every demo, every artifact references the same fictional product so the page reads as one coherent narrative. Your product calibrates to its own context. You describe what it is in /spruce-up, and every Spruce command reasons against that context, not against the example.

A markdown file at the root of your project. It captures your product’s character — what it does, who uses it, what voice it speaks in, how dense it should feel. /spruce-up creates it through a five-minute interview; every other Spruce command reads it before reasoning about your work. The context evolves as the product does.

The Discovery tier writes sibling context files alongside it when you ground the work in users — .personas.md, .jtbd.md, .journeys.md, .scenarios.md. Downstream commands read whichever exist, so the foundation is .spruce.md plus whatever Discovery artifacts you’ve captured. Discovery is optional; the loop reasons better with it in place.

The dimensions themselves are fixed — Typography, Color, Spatial, Component, Motion, UX Writing, UX Patterns. They form the substrate Spruce reasons across. What you customize is how each dimension applies to your product, through the preferences captured in .spruce.md.

Imagery is on the roadmap as an eighth dimension — adding it touches the interview, the template, and the catalog, so it gets its own release rather than sliding in incrementally.

Run any Discovery command and pick Mode B when it asks. Spruce accepts research in whatever shape you have it — paste interview transcripts directly into the conversation, point at a research document in the project, attach survey results, describe usability findings in plain language. The command structures whatever you provide into the artifact format ( .personas.md, .jtbd.md, etc.) and labels each claim with its source.

Confidence labels are the key mechanic. A persona attribute backed by interview data gets labelled research-grounded; one inferred from context gets labelled context-derived; one you know but couldn’t cite gets labelled assumed. Downstream commands ( /audit, /design, /critique) read these labels and weight findings accordingly — a research-grounded conflict gets surfaced as a real tradeoff; a context-derived one gets surfaced as a structured assumption to verify.

Mixed sources are fine. Most projects have a primary persona with deep research and a secondary persona inferred from context — Spruce captures both honestly so the team knows what they actually know.

Five review lenses, each with a different depth and a different question. /detect is fast — a scan for named anti-patterns. /survey is comprehensive — findings with severity tiers and an action plan. /uxreview audits the UX substrate specifically — IA, feedback, forms, state coverage. /critique speaks at the level of feel — does the design have a point of view, does it hedge, is it expressing what it’s trying to express.

/audit is the HCD-grounded counterpart — the only diagnostic that frames findings against named personas and the jobs they’re hiring the product to do. Requires a .personas.md and .jtbd.md in place; without them, /audit recommends running /personas first rather than degrading to generic findings.

For a fuller breakdown, see the catalog.

Spruce works in all five major AI coding tools today: Claude Code, Cursor, Codex CLI, VS Code (with GitHub Copilot), and Gemini CLI. The same npx spruce-skill add command auto-detects which one you’re using and writes the right per-harness bundle. New tools that adopt the Anthropic Agent Skills spec should work via the existing install path with no Spruce changes — that architecture scaled cleanly across the rollout from Cursor to Gemini.

Spruce itself is free under Apache 2.0. You pay for the AI tool you run it through (a Claude Code subscription, or eventually the equivalent for whichever harness you use). Spruce adds no additional cost on top of that.

GitHub issues at iamlaforge/spruce/issues ↗. Bug reports, feature requests, and direction-level questions all welcome — feedback shapes the roadmap.