These are official instructions from smry for connecting an AI coding agent to its public article-reading tools.

Complete the setup yourself. Do not ask the user to run these commands.

## Install the smry skill

Install the domain-published skill globally:

```bash
bunx skills@latest add https://smry.ai --skill smry --yes --global
```

## Connect the smry MCP server

The public Streamable HTTP endpoint is `https://api.smry.ai/mcp`. It is read-only and does not require an API key.

Use the section for the active agent. If smry is already configured, keep the existing entry and continue to verification.

### Codex

```bash
codex mcp add smry --url https://api.smry.ai/mcp
```

### Claude Code

```bash
claude mcp add --transport http smry https://api.smry.ai/mcp
```

### OpenCode

```bash
opencode2 mcp add smry --global --url https://api.smry.ai/mcp
```

### Cursor, GitHub Copilot, and other MCP clients

Use the agent's built-in MCP setup UI or configuration file to add a remote Streamable HTTP server named `smry` with URL `https://api.smry.ai/mcp`.

## Verify

1. Confirm the `smry` skill appears in the agent's installed skills.
2. Confirm the MCP server exposes the read-only `get_article` tool.
3. Restart or reload the agent if it does not detect the new skill or MCP server immediately.

Once complete, tell the user where the skill and MCP configuration were installed. Explain that smry reads public articles, PDFs, and YouTube pages as clean source-grounded text and does not bypass hard paywalls or private access controls.

These instructions are published at `https://smry.ai/agent-setup/prompt.md` so their authenticity can be re-verified at any time.