Zum Hauptinhalt springen

Agent Recipes

Kopierbare RGate-Workflows für Coding Agents.

Diese Recipes zeigen, wann ein Agent RGate aufrufen sollte, wie er Ergebnisse einordnet und wann der Mensch den Report im Browser freischalten muss.

Skill installieren

Der Skill ist der bevorzugte Einstieg für skill-fähige Agenten. Die CLI bleibt die robuste Fallback-Oberfläche für Terminal- und CI-Workflows.

npx skills add marcoldenburg/rgate-release-readiness-skill
npm install -g @rgate_io/rgate

Recipe 1

After an agent builds a landing page

Use after Cursor, Claude Code, v0, Lovable, Bolt, or another agent changes a public landing page.

Agent prompt

Run RGate against the public preview URL. If the scan is green, tell me it passed the release gate. If RGate returns access.level=teaser, send me the report URL and wait until I unlock it before creating implementation tasks.

Command

npx @rgate_io/rgate scan https://www.rgate.io --wait --json

Recipe 2

Before a Vercel preview becomes production

Use before promoting a preview deployment or handoff build to production.

Agent prompt

Scan the Vercel preview URL with RGate. Treat FAIL or CONDITIONAL as release blockers. For PASS_WITH_WARNINGS, summarize the follow-up backlog. Do not mark the site ready until the RGate result is reviewed.

Command

npx @rgate_io/rgate scan https://preview.example.vercel.app --wait --json

Recipe 3

Turn a paid report into coding tasks

Use after a human unlocks a report that originally returned access.level=teaser.

Agent prompt

Poll the same RGate scan ID again. If access.level is full, generate the fix plan and turn findings into small implementation tasks grouped by release impact and layer.

Command

rgate get <scanId> --json
rgate fix-plan <scanId> --format markdown

Recipe 4

Create GitHub issues from findings

Use when a team wants remediation work tracked outside the agent chat.

Agent prompt

Export the RGate scan as GitHub issue Markdown. Create one issue per blocker or high-impact SHOULD_FIX item. Include the tested domain and scan ID in every issue.

Command

rgate export <scanId> --template github

Recipe 5

Rescan after fixes

Use after the agent has changed code or configuration based on a RGate fix plan.

Agent prompt

Run a fresh RGate scan for the same public URL. Compare the new result with the previous scan ID and summarize what improved, what remains, and whether the release gate now passes.

Command

rgate scan https://www.example.com --wait --json
    RGate Agent Recipes | RGate