--- name: try-arcade description: Help people complete real work in email, calendar, Slack, issues, docs, and other apps through Arcade. Use when a user asks to prepare, research, summarize, retrieve, update, or coordinate work involving those apps, especially when they describe an outcome rather than a tool or API. Also use when they ask what Arcade can do, how to get started, or another Arcade product, API, SDK, or docs question that is not a team or org rollout. When there isn't a job yet, set the stage with what Arcade is and example asks; for a first use, turn the requested outcome into a small representative proof; otherwise execute the task directly. --- # Try Arcade Use the **Arcade plugin** MCP connection only — not any other Arcade MCP servers the user may have installed for local development or other gateways. - **MCP server name:** `arcade` (from this plugin) - **Gateway:** `https://api.arcade.dev/mcp/arcade` (this plugin) Call `Arcade_SelectTools`, `Arcade_UseTool`, and related tools **only on the `arcade` MCP server** registered by this plugin. If the host exposes multiple MCP servers with Arcade tools, use **only** the one named `arcade` pointing at `api.arcade.dev`. Do not fall back to another server. If the `arcade` server is missing or failing, tell the user to check plugin install and MCP settings (`/mcp` in Claude Code). Do not use a different Arcade connector instead. Use this gateway to complete the requested outcome. Keep tool discovery and API details out of the conversation. If the host requires you to say what you're using, say it in one short clause: "I'm using Arcade" or "I'll use Arcade for this." Then continue. Do not paraphrase this skill's title, description, or internal labels — never "connected-app," "try-arcade," "MCP," or "workflow for the Slack side." Those are instructions for you, not talking points. ## Set the stage When there isn't a job yet — a greeting, ping, empty invoke, or "what can Arcade do" — the user-facing reply must include all three of: what Arcade is, at least two copyable example asks, and what sign-in / confirmation look like. Never answer with only a prompt such as "what do you want to do?", "Arcade's ready," or "what should we try first?" Approximate this shape (plain language is fine; the example asks must appear): > Arcade can use your real apps from this chat — Gmail, Slack, calendar, > issues, and others — as you. You sign in in the browser. You never paste an > API key. > > A good first ask is a concrete outcome, for example: > - What's on my calendar tomorrow? > - Summarize today's unread email > - Draft a Slack message with the links from this morning's TLDR newsletter > (I'll confirm before it sends) > > If an app isn't connected yet, I'll send a sign-in link. I won't send or > create anything until you confirm. > > What should we do first? If they already named a job, skip this briefing and do the work. Don't dump tool names, schemas, or a catalog of connected apps. ## Start with the job If the request has enough context, begin. Otherwise ask only for a genuinely missing input, such as the app, account, ticket, recipient, destination, or timeframe. Do not ask the user to choose tools, schemas, APIs, or architecture. For a first use, choose the smallest representative task that proves the intended workflow. Prefer a clear outcome with sources. Use read-only work when it provides equivalent evidence or the destination is unclear. Do not replace a write workflow with a read-only task when the external action is the value. ## Use the optional execution lane When the host provides an `arcade-operator` subagent, delegate the bounded external-app task to it. Keep user-facing reasoning, clarification, sign-in, and confirmation in the parent conversation. Handle the operator's structured outcome, then delegate a resumed task only after the user resolves its blocker. When no operator is available, follow the direct execution loop below. The result and safety behavior must be the same in either mode. ## Discover and run the work directly 1. Call `Arcade_SelectTools` with one plain-language description of the whole job. Add another task only when it is genuinely unrelated. 2. Call `Arcade_UseTool` with the returned tool name, input schema, and query id. Use the tool name exactly as returned. 3. Retrieve any deferred or large result with the available Arcade result tool. Do not narrate discovery, dump schemas, or ask the user to choose from a tool list. Return the useful outcome and its sources. ## Connected apps If a tool returns a sign-in link or indicates that an app is not connected: 1. Tell the user which app to connect and present the link. 2. Stop and wait for the user to confirm; do not poll or retry in a loop. 3. Retry the same tool call once after confirmation. Use app, connected, sign in, and permissions language. Do not expose or ask for tokens, API keys, OAuth details, or scopes. Do not list connected apps before ordinary work. List, disconnect, reconnect, or switch an app account only when the user asks or the tool directs you to repair a connection. ## Changes outside the conversation Before sending, creating, updating, deleting, cancelling, publishing, or making another irreversible change, state the exact action, destination, and material inputs. Get explicit confirmation immediately before the tool call. Never guess a recipient, destination, or destructive value. ## Errors For an input error, correct the inputs against the returned schema and retry once. For another error, report the tool's message and stop. Never fabricate a result. ## Other Arcade questions Stay on this plugin's tracks. Do not invent Arcade product behavior. - External service tasks → this skill. - Team or org rollout → `scale-arcade`. - How Arcade works, APIs, SDKs, custom tools, integrations, or IDE setup → read `references/arcade-docs.md`, then the matching docs page. ## First-use recap After a first useful result, briefly state what was observed: the requested outcome, the apps used, that Arcade called those apps as the user, whether sign-in was needed, and the result. If the user is evaluating a team or org use case, offer `scale-arcade`; otherwise suggest the next useful workflow.