dsh-playwright-cli

In DeepSeek Harness agent workflows, running Playwright E2E tests often meant manually typing out long npx playwright commands—probing versions, installing browsers, juggling a dozen CLI flags, then hunting for the HTML report. dsh-playwright-cli packages these scattered steps into four agent-callable tools, eliminating the friction of hand-crafted commands and letting the agent drive the full test lifecycle as naturally as calling a function.

The four tools cover the entire pipeline: playwright_version confirms the CLI is ready; playwright_install downloads browsers (Chromium by default) and scaffolds @playwright/test when needed; playwright_test executes the suite with support for config, project, grep, workers, reporter, trace, retries, and more; playwright_show_report locates the HTML report—registering a loopback route and returning a clickable URL when a web server is available, or falling back to an absolute file path otherwise. Every command rides on the host shell executor, inheriting sandbox policies, environment variables, output truncation, and timeout or abort classification. Non-zero exits are reported rather than thrown as errors, leaving the model to decide the next step.

Ideally suited to frontend or full-stack developers who run E2E tests inside a DSH agent workflow: the plugin lets the agent autonomously close the loop of install-browser, run-tests, and inspect-report with minimal human intervention, while preserving the safety boundary of sandbox approval and permission escalation.