Skills operationalize the Decision Map and core components of Glare.
Inside Glare, AI Skills provide reusable workflows teams use to structure decisions, evaluate direction, compare options, and generate stronger signals across product and design work.
AI accelerated how quickly teams can generate ideas, concepts, prototypes, flows, and experiments. What did not accelerate at the same pace was judgment.
Product and design teams now face a new problem: too much output and not enough structure for evaluating what deserves to move forward. Without structure, organizations accumulate more work, more opinions, and more uncertainty.
AI Skills are repeatable operational systems tied directly to the Decision Map. They help teams move from:
- Ideas to evaluation
- Prompts to workflows
- Feedback to signals
- Signals to decisions
- Decisions to measurable outcomes
Instead of relying on scattered prompts, inconsistent reviews, disconnected workflows, and subjective interpretation, teams apply shared operational patterns that make decision-making clearer and more scalable.
Structure user needs, clarify audiences, build UX metrics, frame stronger questions, and expose assumptions.
Evaluate concepts, structure experiments, analyze findings, generate Design Signals, and interpret user feedback.
Compare directions, benchmark performance, expose tradeoffs, prioritize opportunities, and guide product decisions.
Align stakeholders, operationalize workflows, connect outcomes to business goals, communicate evidence clearly, and scale decision systems across the organization.
| Section | Skills | What it covers |
|---|---|---|
| Get Started | 7 | Onboarding, first signal, AI workflows, glossary |
| Design Review | 12 | SIGNAL framework (Surface, Identify, Ground, Navigate, Align, Lock), scoring, coaching |
| Design Signals | 5 | Signal anatomy, types (Need/Use/Prefer/Adopt), quality, capturing |
| Decision Map | 21 | Define / Measure / Focus / Lead (4 sub-skills each + parent) |
| Design Assessment | 9 | 5 org-maturity dimensions, scoring, pattern reading |
| UX Metrics | 1 | 4 metric types (Attitudinal, Behavioral, Performance, Intelligence) |
| Navigation | 1 | glare-router cross-section meta-skill |
/plugin marketplace add zurb/glare-skills
/plugin install glare@glare-marketplace
Then run /reload-plugins. Skills load under the plugin's name, as glare:measure-hunches, glare:define-ux-metrics and so on, and auto-trigger when your conversation matches their domain.
Ray stores the Decision Map as records: situations, user needs, audiences, concepts, hunches, questions, findings, initiatives, methods, bets, business goals, workflows, results and metrics. With the Ray MCP connected, the skills that produce one of those offer to save it to a Ray project, and glare:decision-map and glare:master read the project before advising. Each of those skills has a "Saving to Ray" section naming the tool and the links. Without Ray, the sections are skipped and the skills work the same.
npx skills add zurb/glare-skills
Since 4.0.0 the skill folders no longer carry a glare- prefix, so a flat install names them measure-hunches rather than glare-measure-hunches, and they can collide with other skills of the same name. Prefer the plugin install where your tool supports it.
Download a skill file and drop it into your LLM chat. The portable prompt versions live in glare-prompts-flat/ -- one .md file per skill, plus a master prompt covering the entire framework.
Each skill is a self-contained prompt module with:
- SKILL.md -- YAML frontmatter with trigger conditions + step-by-step coaching instructions
- reference.md -- Compressed source material (worksheets, tables, worked examples)
You don't invoke skills by name. Claude reads each skill's description and loads the right one when your message matches. Examples:
| Say this | Loads this |
|---|---|
| "let's pick metrics for checkout" | glare:define-ux-metrics |
| "we need to benchmark v1 vs v2" | glare:focus-comparing |
| "how do I tell sales this is working?" | glare:lead-workflows |
| "we believe simplifying onboarding will improve activation" | glare:measure-hunches |
| "I'm stuck and don't know where to start" | glare-router |
A typical workflow might frame a user problem, generate multiple concepts, structure testing criteria, compare directions, analyze UX metrics, identify stronger signals, guide a recommendation, and prepare a decision review. Each skill focuses on a specific part. Together, they create systems for evaluation, comparison, interpretation, prioritization, alignment, and decision-making.
For teams that want a focused install rather than the full set:
| Package | Skills | Use case |
|---|---|---|
| Core Decision Map | 21 | Just the Decision Map cluster + router |
| Design Review | 12 | Just the SIGNAL framework + scoring/rubric/coach |
Most organizations no longer struggle to generate ideas. They struggle to evaluate what matters, maintain alignment, reduce uncertainty, move with confidence, and preserve momentum.
AI Skills help teams operationalize decision-making itself. Over time, Skills become reusable systems organizations rely on to create clearer workflows, stronger signals, faster alignment, more measurable decisions, and scalable product and design operations.
This is where AI becomes more than output generation. It becomes part of the operational system teams use to guide direction, strengthen confidence, and make better product and design decisions.
2.1.0 (2026-05-08) -- Full 56-skill set restored. See CHANGELOG.md for complete history.
Open source -- github.com/zurb/glare-skills