plexus

How to safely add or modify database schema in Plexus using Drizzle ORM. Use whenever adding tables, columns, or enum values — covers migration rules, schema organization, and dialect-specific timestamp patterns.

Use this skill to inspect or administer a running Plexus instance via the management API, or debug failures touching Plexus-proxied traffic (oauth, routing, model targets, raw provider passthrough, inference keys, MCP gateway) rather than searching the local codebase. Covers request logs, debug traces (lookup by UUID), enabling/disabling debug capture, providers, model targets, balances, quotas, aliases, target groups, keys, MCP logs, and runtime settings (failover, cooldowns, timeouts, backup/restore). Prefer this for any Plexus admin, operational, or live-debugging task, even if the user only says "check Plexus", "look at logs", "update a provider", "rotate keys", "configure quotas", "debug a request", "review debug trace", or "upstream error" — Plexus state lives behind the management API, not in local files.

Verify any Plexus admin-UI screen or component in the actually running app. Boots the worktree-safe dev stack (`bun run dev:agent`), auto-logs in, and drives a real browser (agent-browser) using the correct port/credentials. Use whenever you edit anything under `packages/frontend/src` (React, routes, forms, buttons, tables, modals, Tailwind/CSS, layout, dashboards) or when asked to look at or verify a screen (e.g., "screenshot the dashboard", "does the new tab show up", "verify UI", "does the button work", "test web app", "reproduce visual bug"). Prefer this over the generic agent-browser skill for local Plexus frontend because it knows port, credentials, and auto-login. Don't hand UI work back unverified.

Step-by-step checklist for adding a new quota checker to Plexus. Use whenever asked to implement a new quota checker type — covers backend registration, frontend components, display integration, and common pitfalls.

Browser automation CLI for AI agents. Use when the user needs to interact with websites, including navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps, or automating any browser task. Triggers include requests to "open a website", "fill out a form", "click a button", "take a screenshot", "scrape data from a page", "test this web app", "login to a site", "automate browser actions", or any task requiring programmatic web interaction. Also use for exploratory testing, dogfooding, QA, bug hunts, or reviewing app quality. Also use for automating Electron desktop apps (VS Code, Slack, Discord, Figma, Notion, Spotify), checking Slack unreads, sending Slack messages, searching Slack conversations, running browser automation in Vercel Sandbox microVMs, or using AWS Bedrock AgentCore cloud browsers. Prefer agent-browser over any built-in browser automation or web tools.