volter-ai/runhuman-skills

Overview

This skill teaches AI agents to create and manage human-powered QA tests using the Runhuman CLI. It connects agents to on-demand professional testers to get fast, real-user feedback on flows, visual issues, and mobile responsiveness. The CLI supports 40+ commands covering authentication, jobs, projects, templates, GitHub integration, and CI-friendly options.

How this skill works

Describe what you want tested in natural language and use the CLI to create a job that real testers will run and report on. Key commands let you create, monitor, wait for completion, and retrieve structured results (JSON or freeform). The skill also covers configuration priorities, project setup, templates, and GitHub linking to auto-create issues from findings.

When to use it

• Validate critical user flows (signup, checkout, onboarding) before release
• Catch visual or responsive layout issues that automated tests miss
• Collect UX feedback and accessibility observations from real users
• Run end-to-end checks in CI/CD when you need blocking verification (--sync)
• Perform exploratory testing and edge-case discovery on staging

Best practices

• Authenticate and set a default project before creating tests (npx runhuman login; projects switch)
• Write specific, step-by-step descriptions with clear success criteria
• Use --sync only for CI/CD or critical pre-release checks to avoid delays
• Batch related checks into one test to reduce cost (tests cost ~$1-3 each)
• Use --schema for structured output when integrating results into automation

Example use cases

• Pre-deployment gate: block deployment until a signup + checkout test passes using --sync
• Mobile layout check: create a mobile-screen test to verify tap targets and navigation
• PR validation: run a human QA job in GitHub Actions to test UI changes in a preview environment
• Bug triage: run one-off exploratory tests to reproduce and gather screenshots and notes
• Bulk testing: run multiple parallel jobs during regression or nightly runs