ridgeline

Validate that the repo and a build's directory are ready for `ridgeline build`. Runs `npm run check`, parses `.check/summary.json`, verifies the planner has produced phases, checks node_module deps, git state, and spec-vs-repo consistency. Identifies anything that would hang up a build agent or burn budget on retries — missing system deps, broken tests, stale phase artifacts, spec referents that don't exist. Writes build-check.md with status ready, needs-attention, or broken. Use when the user runs `/build-check [build-name]` or asks to validate readiness before `ridgeline build`.

Read a build's build-check.md and fix repo issues so `ridgeline build` can run cleanly. Plans fixes, dispatches parallel subagents grouped by disjoint file sets, re-runs `npm run check` and `/build-check`, and loops until status reaches ready or a 3-iteration cap is hit. Use when the user runs `/build-fix [build-name]` or asks to repair the repo after `/build-check` flagged issues.

Validate that a build's spec.md is ready for the ridgeline planner. Checks required sections, internal consistency, cross-file alignment with constraints/taste/design, and feasibility against the repo. Writes spec-check.md with status ready, needs-attention, or broken. Use when the user runs `/spec-check [build-name]` or asks to validate a build's spec before planning. Run before `ridgeline plan` to catch spec issues that would cause the planner to hallucinate phases.

Read a build's spec-check.md and fix the spec/constraints/taste/design files so the spec is ready for the ridgeline planner. Plans fixes, dispatches one or more fix agents, re-runs `/spec-check`, and loops until the status reaches ready or a 3-iteration cap is hit. Use when the user runs `/spec-fix [build-name]` or asks to repair a build's spec after `/spec-check` flagged issues.

Convert an arbitrary spec, PRD, RFC, or design doc (or a directory of related docs) into a ridgeline-ready trio of `spec.md` + `constraints.md` + `taste.md` (and optional `design.md`) saved to a target directory. Use this skill whenever the user wants to feed a freeform specification into ridgeline, asks to "ridgeline-ify" or "split" a spec, mentions converting a PRD/RFC/design doc into ridgeline inputs, or requests producing constraints/taste/spec files from a source document or document folder. Do NOT trigger for editing an existing constraints.md/taste.md/spec.md — only for the initial extraction from source material.

Bump the project version in `package.json` via scripts/bump-version.mjs, summarize every commit since the last release into a new CHANGELOG.md section, commit with a `vX.Y.Z` message, and push an annotated tag. Use when cutting a release.

Validate and cross-compile GLSL, WGSL, and SPIR-V shaders using naga. Use when writing shaders, checking shader compilation, debugging shader errors, converting between shader languages, or verifying WebGL/WebGPU shader code.

Capture annotated browser screenshots with numbered element labels for visual verification. Use when building or reviewing web UIs, verifying responsive layouts, checking visual output of canvas/WebGL content, or inspecting rendered pages. Trigger when asked to screenshot, verify layout, check rendering, or visually inspect a running web app.

Capture rendered canvas and WebGL frames from browser-based games and visual applications. Use when verifying canvas rendering, checking WebGL output, capturing game screenshots, or validating visual output from PixiJS, Phaser, Three.js, or raw canvas apps.

Run WCAG 2.1 AA accessibility checks using axe-core. Use when verifying accessibility compliance, checking contrast ratios, validating ARIA usage, auditing keyboard navigation, or reviewing landmark structure.

Analyze CSS for design system drift, specificity issues, and bloat using Project Wallace. Use when reviewing CSS quality, checking for unused rules, auditing color and font consistency, or detecting specificity problems in stylesheets.

Compare screenshots pixel-by-pixel against reference images to detect visual regressions. Use when checking for unintended visual changes, comparing before/after screenshots, or validating that UI changes match expectations.

Run Lighthouse audits for performance, accessibility, and best practices scoring. Use when checking page performance, running quality audits, measuring Core Web Vitals, or getting a quantitative quality score for a web page.