> ## Documentation Index
> Fetch the complete documentation index at: https://docs.anchorage.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Ready-made prompts

> Start common Anchorage Digital integrations from copy-paste AI prompts.

Copy these prompts into your AI assistant to scaffold common integrations. For the best results, connect the [docs MCP server](/knowledge-base/build-with-ai/connect-ai-assistant) first so the assistant works from the current documentation.

## Integration prompts

<Prompt description="Scaffold an API client" actions={["copy"]}>
  You are helping build an integration with the Anchorage Digital REST API (base URL [https://api.anchorage.com/v2](https://api.anchorage.com/v2)). Using the Anchorage Digital documentation, scaffold a typed API client that authenticates with the `Api-Access-Key` header and signs sensitive requests with an Ed25519 `Api-Signature` and `Api-Timestamp`. Cite the specific documentation pages you rely on, and confirm the current endpoint paths from the docs rather than assuming them.
</Prompt>

<Prompt description="Set up a withdrawal flow" actions={["copy"]}>
  Using the Anchorage Digital documentation, outline and implement the steps to withdraw a digital asset: creating a permission group, generating an API key, adding a trusted destination, and submitting the withdrawal. Call out every step that requires quorum approval, and do not bypass human approval. Verify each endpoint against the docs before writing code.
</Prompt>

<Prompt description="Configure webhooks" actions={["copy"]}>
  Using the Anchorage Digital documentation, help me subscribe to and verify webhooks for transfer and deposit events. Show how to validate webhook signatures and how to reconcile each event with the read APIs.
</Prompt>

<Prompt description="Convert stablecoins" actions={["copy"]}>
  Using the Anchorage Digital documentation, implement a stablecoin mint, redeem, and bridge with the conversion endpoint, and explain how to track each conversion to completion. Confirm the exact asset-type strings from the docs.
</Prompt>

## Keep prompts available in your tool

To reuse guidance across a project, save it in your tool's rules or instructions file:

| Tool           | Where to add project instructions               |
| :------------- | :---------------------------------------------- |
| Claude Code    | `CLAUDE.md` or `AGENTS.md` in your project root |
| Cursor         | Project rules in `.cursor/rules`                |
| GitHub Copilot | `.github/copilot-instructions.md`               |
| Windsurf       | Workspace rules in `.windsurf/rules`            |

<Note>
  Exact file names change as tools evolve. Confirm the current path in your AI tool's documentation.
</Note>

## Write your own

Effective prompts share a few habits: point the assistant at our docs or the MCP server, name the exact task, ask it to cite the pages it used, and tell it to verify endpoints and asset types against the docs rather than relying on memory.

<Tip>
  Always review AI-generated code before running it, and test in the [sandbox](/knowledge-base/platform/developers/sandbox/sandbox-overview) before production. See [AI security and guardrails](/knowledge-base/build-with-ai/security).
</Tip>