Skip to main content

Agent Skill

The qas-cli repository ships a SKILL.md — a condensed reference written for AI coding agents (Claude Code, Cursor, and any other agent that supports the skills format). Once the skill is registered, the agent can drive QA Sphere on your behalf — listing projects, authoring test cases, opening and closing runs, posting results, and pulling reports — without you scripting any of it.

Install the skill​

npx skills add Hypersequent/qas-cli

The command writes the skill into your agent's local skill registry. Re-run it to update the skill in place.

Per-agent setup

Each agent has its own way of activating registered skills. See the Vercel skills registry for the current per-agent install instructions and supported runtimes.

Prerequisites​

  • qas-cli installed (npm install -g qas-cli or available via npx qas-cli). See Overview.
  • Authenticated session (qasphere auth login) or QAS_TOKEN + QAS_URL configured. See Auth.

The skill reads the same qasphere <command> -h output a human would, so it stays current with the installed CLI — there's nothing for you to keep in sync.

What the agent can do​

With the skill loaded, you can describe the outcome you want in natural language and the agent composes the right qasphere api ... calls. The CLI prints JSON; the agent renders it however the conversation needs — Markdown tables, matplotlib/Plotly charts, SVG, HTML, CSV exports.

A few examples that work out of the box:

  • "Plot the pass rate of the last 20 closed runs in project PRJ as a line chart." → the agent calls qasphere api runs list --project-code PRJ --closed true, computes passed / all per run from statusCounts, renders a chart.
  • "Which folders in PRJ have the most failing tests this week?" → combines runs list, runs test-cases list, and folders list to aggregate failures by folder.
  • "Generate a Markdown summary of run 42 with the pass/fail breakdown and a list of failing test case titles." → calls runs test-cases list --project-code PRJ --run-id 42 and formats the output.
  • "Create a folder structure for the Authentication area with Login, OAuth, and MFA subfolders, then scaffold five test cases under Login." → calls folders bulk-create followed by test-cases create.

You don't need to pre-script anything — describe the report or change you want, and let the agent compose the API calls.

Further reading​