> For the complete documentation index, see [llms.txt](https://docs.amply.tools/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.amply.tools/getting-started/ai-assisted-integration.md). # AI-assisted integration There are two ways to get Amply into your app. You can wire it **by hand** — add the SDK, forward your events, set the attributes you target on (that's the [Developer Guide](/developer-guide/installation.md)). Or you can let an **AI coding assistant do it for you** — this page. Connect Amply to your assistant and it does the work: it reads your app and sets the SDK up, and it turns a plain-language request into a real campaign. You review and approve; the assistant does the legwork. This doesn't replace the dashboard or the SDK — it's a faster door to both. Developers can still integrate by hand and PMs can still build campaigns in the dashboard. The AI-assisted path is for teams who'd rather describe the outcome and review the result than do every step themselves. ## Set it up Two pieces, installed once. You need [Node](https://nodejs.org) and an MCP-capable AI assistant (Claude Code, the Codex CLI, or any client that supports MCP). **1. The Amply MCP** — connects your assistant to your Amply account. * **Claude Code:** `claude mcp add amply -- npx -y @amplytools/amply-mcp` * **Codex CLI:** `codex mcp add amply -- npx -y @amplytools/amply-mcp` * **Other clients:** point them at the `@amplytools/amply-mcp` package over stdio. On first use the assistant signs you in to Amply; after that it can read your projects and author campaigns on your behalf. **2. The integration skill (`amply-integration`)** — an open-source playbook that tells the assistant *how* to add Amply well: detect the platform and package manager, audit the analytics the app already has, wire the SDK through a single thin layer, set up the custom properties you'll target on, and verify the build itself. Install it whichever way suits your assistant: {% tabs %} {% tab title="Any assistant (skills CLI)" %} Works with Claude Code, the Codex CLI, and any other host the [`skills` CLI](https://skills.sh) supports: ```bash npx skills add amply-tools/sdk-skill ``` Add `-g` to install for your user instead of the current project. Target a specific assistant with `-a` (e.g. `-a claude-code` or `-a codex`); run `npx skills --help` for the full list of supported assistants. {% endtab %} {% tab title="Claude Code (plugin)" %} Claude Code can install the skill as a plugin from Amply's plugin catalog. Run both commands once, inside Claude Code: ``` /plugin marketplace add amply-tools/claude-plugins /plugin install amply-integration@amply ``` The first command **adds Amply's plugin catalog** — it installs nothing on its own. The second command **installs the skill** from that catalog. (`amply` is the catalog name; `amply-integration` is the skill.) {% endtab %} {% endtabs %} Either path installs the same skill — pick one. Source: [github.com/amply-tools/sdk-skill](https://github.com/amply-tools/sdk-skill) The MCP is the *connection* to your account; the skill is the *know-how* for a clean integration. With both in place you describe the outcome and review the result. ## Integration, the AI way Integrating an analytics-style SDK is usually a checklist a developer works through by hand: install it, find the events the app already fires, forward the ones that matter, and set the user attributes you'll target on. It's mechanical, easy to get subtly wrong, and slow to repeat across apps. > When a developer connects the Amply MCP to their AI coding assistant and asks it to add Amply, the assistant audits the app's existing analytics, installs and initializes the SDK, forwards the events Amply needs for targeting, and sets up custom properties — as a reviewable change in the codebase, not a black box. The developer reads the diff and merges. The result: the same integration a developer would do, produced in one pass and handed back for review — so the judgment stays with you and the typing doesn't. ## Running campaigns by describing them Campaign work normally means a person in the dashboard translating an idea into targeting, a trigger, and an action. MCP lets you state the idea and get the campaign. > When a PM asks the assistant to "show a rewarded-ad gate before export, for free users, once per session," the MCP turns that into a campaign in Amply — audience, trigger, and action wired — for you to review and activate. Editing a campaign or checking how one is set up works the same way: ask, review, approve. Common shapes are a single request away — a deeplink when a user's status changes (for example, trial → expired), a prompt after a positive moment, or a paywall on the Nth session. The assistant picks the matching campaign shape, fills in your audience and trigger, and leaves it in Draft for you to review. You stay in control of what goes live: the assistant proposes, you approve in the dashboard. It compresses the path from idea to live campaign; it doesn't take the decision away from you. ## When to reach for it * **Integrating, or adding Amply to another app** — let the assistant do the SDK pass and hand you a diff. * **Spinning up or adjusting campaigns quickly** — describe the change instead of clicking through the form, then review. * **You'd still use the dashboard** for a careful look before activating, and for anything you'd rather see laid out visually. ## Related * [Quickstart](/getting-started/quickstart.md) — the by-hand path from zero to a working integration * [User Guide](/user-guide/overview.md) — building and managing campaigns in the dashboard * [Developer Guide](/developer-guide/installation.md) — integrating the SDK by hand * [amply-integration skill](https://github.com/amply-tools/sdk-skill) — the open-source playbook the assistant follows --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter: ``` GET https://docs.amply.tools/getting-started/ai-assisted-integration.md?ask= ``` The question should be specific, self-contained, and written in natural language. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.