gem-hunter

Run QA Wolf flows from your terminal or in CI.

The CLI runs and manages flows locally. Flow creation, AI-powered test generation, managed cloud execution, and team collaboration are part of the full QA Wolf platform. See docs.qawolf.com for the full docs, or qawolf.com to get started.

npm install -g @qawolf/cli
# or: pnpm add -g @qawolf/cli
# or: yarn global add @qawolf/cli
# or: bun add -g @qawolf/cli

Try it without installing:

npx @qawolf/cli --help

Supported Node versions: Node 24.14.1+.

Precompiled binaries are attached to each GitHub Release — no Node.js required: qawolf-linux-x64, qawolf-linux-arm64, qawolf-darwin-x64, qawolf-darwin-arm64, and qawolf-windows-x64.exe.

curl -fsSL -o qawolf https://github.com/qawolf/cli/releases/latest/download/qawolf-darwin-arm64
chmod +x qawolf && ./qawolf --help

You need a QA Wolf account — sign up at qawolf.com. The <env-id> comes from the QA Wolf dashboard under Settings → Environments.

qawolf auth login                  # or set QAWOLF_API_KEY for CI
qawolf flows run --env <env-id>

qawolf flows run --env runs your team's flows from the local .qawolf/<env> cache (a per-environment copy of your flows on disk), pulling them first only if they are not already cached locally and installing the runtime dependencies they need. To refresh the local cache, run qawolf flows pull --env <env-id>; to author flows locally without the platform, run qawolf init first.

The examples/ directory contains sample flows you can run directly:

qawolf install   # one-time: install the browser runtime
qawolf flows run examples/example.flow.ts

Run any command with --help for its flags and options. qawolf help ref <command> prints the help of a command and of every command under it, as Markdown.

Start with a run ID, pick an attempt ID from its failed flows, then narrow the evidence only when the summary points to it:

qawolf run get --run-id <run-id> --flow-statuses failed
qawolf attempt investigate --attempt-id <attempt-id>
qawolf attempt inspect --attempt-id <attempt-id> --timeline --limit 20
qawolf attempt inspect --attempt-id <attempt-id> --request <request-id>

attempt inspect requires exactly one of --action, --timeline, --network, --request, --snapshot, --screenshot, --console, or --log. Use --evidence-id with --console or --log to return exactly that one evidence record; the other filters are listed in qawolf attempt inspect --help.

Snapshot HTML can be returned directly or written to a new file. Screenshots require a file. The CLI refuses to overwrite an existing path:

qawolf attempt inspect --attempt-id <attempt-id> --snapshot <snapshot-id> --format html
qawolf attempt inspect --attempt-id <attempt-id> --snapshot <snapshot-id> --format html --output-file snapshot.html
qawolf attempt inspect --attempt-id <attempt-id> --screenshot <screenshot-id> --output-file screenshot.jpeg

The trace and logs are downloaded from the signed links qawolf attempt get returns and parsed on your machine. Downloads, output text, HTML, request bodies, and log retention are bounded; incomplete, truncated, malformed, and artifact status fields describe evidence that could not be read in full.

This repository ships a qawolf-cli Agent Skill that follows the open Agent Skills specification. It works with compatible harnesses such as Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, and OpenCode.

Install the CLI first, then use the cross-harness skills installer to install the skill globally:

npm install -g @qawolf/cli
npx skills add qawolf/cli --skill qawolf-cli --global

The installer detects supported harnesses and prompts you to select where to install the skill. To target specific harnesses non-interactively:

npx skills add qawolf/cli \
  --skill qawolf-cli \
  --global \
  --agent claude-code \
  --agent codex \
  --yes

Omit --global to install the skill only for the current project. The skill also ships in the npm package for consumers that manage skill files directly.

The skill is generated from its source template and the command tree (bun run generate), and kept in sync by the test suite.

Drive interactive runners from TypeScript instead of spawning commands. Every verb names the runner it addresses, so nothing is launched or billed implicitly.

import { createRunnerSdk } from "@qawolf/cli/runner-sdk";

const runner = createRunnerSdk({ apiKey: process.env.QAWOLF_API_KEY });

await runner.launch({ id: "agent-1", runnerFamily: "default" });

const submitted = await runner.run({
  entryPointPath: "src/flows/smoke.flow.ts",
  environment: "ambient",
  runnerId: "agent-1",
  selection: "whole-flow",
});
if (submitted.ok) console.log(submitted.value.runId, submitted.value.fileSync);

ok: false reports a request failure, including transport, auth, or local validation. A runner that refused the request comes back as ok: true with an outcome and failureReason from the published contract, so a caller can switch on it exhaustively. Most verbs expose value.outcome; recording controls expose value.result.outcome and a history value.url.

Control video recording on a Playwright runner after a flow has started its screen:

const recordingId = crypto.randomUUID();
const recording = await runner.record({
  runnerId: "agent-1",
  command: { action: "start", recordingId },
});
if (recording.ok && recording.value.result.outcome === "success") {
  // Drive the browser, then stop this specific capture.
  await runner.record({
    runnerId: "agent-1",
    command: { action: "stop", recordingId },
  });
}

// History remains readable after the runner terminates.
const history = await runner.listRecordings({ runnerId: "agent-1" });
if (history.ok)
  console.log(history.value.recordings, history.value.nextPageToken);

record also accepts { action: "status" } and { action: "auto", enabled: true } (or false). Keep the UUID when retrying start or stop. listRecordings accepts optional recordingId and pageToken filters and returns one page, with stable platform URLs and expiring video URLs. The equivalent CLI commands are qawolf runner record start|stop|status|auto and qawolf runner list-recordings; see the runner reference.

• Commands — full command and flag reference
• Configuration — qawolf.config.ts fields
• Environment variables
• Exit codes
• Troubleshooting
• Known issues — current limitations and workarounds

See CONTRIBUTING.md for development setup. To report a bug or request a feature, open an issue on GitHub.

To report a vulnerability, see SECURITY.md.

Apache-2.0