GitHub

Slash Commands

Slash commands close the development loop inside the agent session. Building, auditing, and verifying performance happen without leaving the conversation — and every audit is checked against the thresholds you have declared in config.

Available commands

CommandWhat it does
/pulse-devStarts (or restarts) the development server. The server watches for file changes and reloads automatically.
/pulse-stopStops the running development server.
/pulse-buildRuns a production build. Bundles all specs via esbuild into public/dist/ with content-hashed filenames.
/pulse-startStarts the production server against the built output. Used to verify production behaviour before deploying.
/verifyRuns the full verification loop: validate, screenshot, Lighthouse desktop + mobile (pass bar: 100 on Accessibility, Best Practices, and SEO), a mobile layout check, a performance trace, and pulse_review — then writes the verification stamp. /verify --quick skips Lighthouse and the perf trace for fast mid-build checkpoints.

Using commands

Commands are typed directly into the agent chat:

/pulse-dev
/verify

The agent executes the relevant CLI steps and reports back with results, including whether any Lighthouse score fell short of the pass bar.

/verify runs Lighthouse against a full production build (pulse_build, then the production server on port 3001) — never against the dev server. Development builds are unminified and report misleading scores.

Plain language prompts

Slash commands cover the most common operations. For everything else, describe the goal — the agent handles the implementation within Pulse's spec structure:

"Create a blog index page that fetches posts from an API"
"Add email validation to the contact form"
"Build a checkout flow with a Stripe payment step"
"Add a guard to the dashboard so unauthenticated users are redirected to /login"

The agent produces spec files that conform to Pulse's structure — the framework enforces correctness, so there is no manual wiring to verify.