Production-oriented agent skills for VertaaUX accessibility, UX, and conversion audits.

This repository packages VertaaUX guidance in the open Agent Skills format so coding agents can run audits, interpret findings, and turn results into CI gates or remediation plans without inventing unsupported commands.

Quick Start

Install the published skills package:

npx skills add VertaaUX/agent-skills

Preview the packaged skills before installing:

npx skills add VertaaUX/agent-skills --list

Install only the VertaaUX skill:

npx skills add VertaaUX/agent-skills --skill vertaaux

Run a minimal VertaaUX workflow:

npm install -g @vertaaux/cli
vertaa login
vertaa audit https://example.com --profile quick-ux --wait --format json > audit.json
cat audit.json | vertaa triage

What This Repo Gives You

• A production-ready vertaaux skill for URL audits, WCAG investigations, CI gates, and agent-driven remediation workflows
• Reference documents for audit profiles, CLI workflows, CI/CD setup, SDK/API usage, and reusable playbooks
• Guardrails that keep agents aligned with real VertaaUX surfaces instead of hallucinated flags or parameters
• Composition contracts for handing off VertaaUX work into adjacent review or architecture skills

Available Skill

vertaaux

Run and operationalize VertaaUX audits across CLI, CI/CD, SDK, API, and MCP.

Use it when the user needs to:

• audit a live URL for UX, accessibility, or conversion issues
• investigate WCAG findings or accessibility regressions
• set up CI or PR quality gates with thresholds and baselines
• compare audit runs and explain score deltas
• generate triage, fix plans, patch reviews, or team playbooks from audit output

What It Covers

• Audit profiles: built-in and custom profile definitions, selection decision tree, and category filtering
• Task recipes: deterministic step sequences for accessibility audits, monitoring, remediation, and competitive review
• Verification and drift control: guardrails that prevent the agent from inventing flags, parameters, or commands the CLI/API doesn't actually expose
• Skill composition contracts: explicit input/output/handoff conventions so vertaaux can chain into a11y-review, create-analyzer, and architecture-review without guesswork
• CLI workflows for one-off audits and AI follow-up commands
• CI/CD setup for score thresholds, baselines, and regression detection
• SDK, API, and MCP integration for automated workflows and agent-driven tooling

Typical Outcomes

• Fast audit selection with documented profiles such as quick-ux, wcag-aa, and ci-gate
• Deterministic follow-up flows for explanation, triage, comparison, and fix planning
• Operational CI guidance for score thresholds, baselines, and regression detection
• Programmatic integration patterns for SDK, API, webhooks, and MCP-driven agents

Repository Layout

• skills/vertaaux/SKILL.md - Main skill instructions and routing guidance
• skills/vertaaux/references/audit-profiles.md - Built-in profiles, custom profile schema, and profile selection guidance
• skills/vertaaux/references/cli-workflows.md - Command reference and piping patterns
• skills/vertaaux/references/cicd-setup.md - CI/CD and GitHub Actions examples
• skills/vertaaux/references/sdk-api.md - SDK, API, webhook, and MCP details
• skills/vertaaux/references/use-cases.md - Task recipes and step-by-step workflow playbooks
• skills/vertaaux/references/skill-contracts.md - Skill composition contracts for chaining vertaaux with a11y-review, create-analyzer, and architecture-review

Compatibility

This repository is designed for tools that support the Agent Skills format.

Common hosts include:

If your environment can install Agent Skills from GitHub, this repository should fit directly into that workflow.

Project Standards

• Changes should preserve documented VertaaUX behavior and avoid inventing unsupported CLI flags or API fields.
• Documentation updates should keep examples executable and relative links valid.
• Repo hygiene checks run in GitHub Actions on pull requests and pushes to main.

See CONTRIBUTING.md for contribution expectations and SECURITY.md for vulnerability reporting guidance.

License

MIT