> ## Documentation Index
> Fetch the complete documentation index at: https://docs.meld.io/llms.txt
> Use this file to discover all available pages before exploring further.
# Build with AI
> Meld's Stablecoins & Digital Assets docs are AI-friendly. Plug them into Claude, Cursor, ChatGPT, or any LLM tool to ship your crypto integration faster.
The Meld Stablecoins documentation is published in an agent-ready format so coding assistants and AI agents can read, search, and reason over it directly — without copy-pasting pages or relying on stale training data.
This page covers four ways to bring Meld docs into your AI workflow:
Connect Claude, Cursor, and other tools to a live, searchable index of the docs.
Drop-in context files that summarize or inline the entire docs site.
Get any page as clean Markdown by appending `.md` to its URL.
Machine-readable capability file for agents that act on Meld's APIs.
All endpoints below serve the full Meld documentation, including the Stablecoins & Digital Assets, Meld API, and API reference sections.
## MCP server
The Meld docs are exposed as a hosted [Model Context Protocol](https://modelcontextprotocol.io) server. Once connected, your AI tool can search the docs and read full pages on demand instead of guessing.
**Server URL**
```
https://docs.meld.io/mcp
```
**Tools the server exposes**
* `search_meld` — semantic search across every page; returns titles, snippets, and links.
* `query_docs_filesystem_meld` — shell-style read access (`ls`, `tree`, `cat`, `head`, `rg`) over the docs as a virtual filesystem, including the OpenAPI specs.
### Add it to your tool
Run from your project:
```bash theme={null}
claude mcp add --transport http meld-docs https://docs.meld.io/mcp
```
Then in a session, ask: *"Use the meld-docs MCP to find the White-Label API quickstart and summarize the request payload."*
Add to `claude_desktop_config.json`:
```json theme={null}
{
"mcpServers": {
"meld-docs": {
"type": "http",
"url": "https://docs.meld.io/mcp"
}
}
}
```
Restart Claude Desktop. The `meld-docs` tools will appear in the tool picker.
Add to `~/.cursor/mcp.json` (global) or `.cursor/mcp.json` in your project:
```json theme={null}
{
"mcpServers": {
"meld-docs": {
"url": "https://docs.meld.io/mcp"
}
}
}
```
Add to your workspace `.vscode/mcp.json`:
```json theme={null}
{
"servers": {
"meld-docs": {
"type": "http",
"url": "https://docs.meld.io/mcp"
}
}
}
```
Open a chat and attach the Meld docs MCP under **Connectors → Add connector** (available on Plus, Pro, Business, and Enterprise). Use:
```
https://docs.meld.io/mcp
```
No auth required.
A discovery endpoint is published at `https://docs.meld.io/.well-known/mcp` so MCP-aware tools can find the server automatically.
## llms.txt
The docs are also published as flat text files designed for direct LLM consumption — useful when you want to drop the whole site into a context window, a RAG pipeline, or a fine-tuning corpus.
| File | Use it when |
| -------------------------------------------------------------------------- | ------------------------------------------------------------ |
| [`https://docs.meld.io/llms.txt`](https://docs.meld.io/llms.txt) | You want a structured index of every page with descriptions. |
| [`https://docs.meld.io/llms-full.txt`](https://docs.meld.io/llms-full.txt) | You want the entire docs site inlined as one Markdown file. |
**Example — load the full docs into a Claude API request:**
```python theme={null}
import anthropic, urllib.request
docs = urllib.request.urlopen("https://docs.meld.io/llms-full.txt").read().decode()
client = anthropic.Anthropic()
msg = client.messages.create(
model="claude-opus-4-8",
max_tokens=1024,
system=[
{"type": "text", "text": "You are a Meld integration assistant."},
{"type": "text", "text": docs, "cache_control": {"type": "ephemeral"}},
],
messages=[{"role": "user", "content": "How do I create a crypto quote for a USD → USDC buy?"}],
)
print(next((b.text for b in msg.content if b.type == "text"), ""))
```
Use prompt caching (as shown above) when you call repeatedly — `llms-full.txt` is large, and caching cuts cost and latency dramatically.
## Markdown export
Any page on `docs.meld.io` can be fetched as raw Markdown — no HTML, no nav chrome — by appending `.md` to its URL.
**Examples**
```
https://docs.meld.io/docs/stablecoins/crypto-overview_/quickstart.md
https://docs.meld.io/docs/stablecoins/white-label-api-integration/whitelabel-quickstart.md
https://docs.meld.io/api-reference/crypto/retail-ramp/crypto-quote-get.md
```
You can also:
* Send `Accept: text/markdown` on any docs request to get the Markdown variant.
* Press Cmd+C (Ctrl+C on Windows) on any page to copy it as Markdown.
* Use the **Copy as Markdown** / **View as Markdown** options in the page's contextual menu.
API reference pages include the full OpenAPI operation in the Markdown export, which is what most coding agents need to generate a correct request.
## skill.md
For agents that *act* on Meld (not just read about it), the docs publish an [agentskills.io](https://agentskills.io/specification)–compatible capability file:
```
https://docs.meld.io/skill.md
```
It describes what an agent can accomplish with the Meld API, the inputs each capability needs, and the constraints that apply. MCP-connected agents discover it automatically as a resource on the server above; you can also add it manually:
```bash theme={null}
npx skills add https://docs.meld.io
```
## Recommended setup for crypto integrations
Add `https://docs.meld.io/mcp` to your editor or chat tool. This is the single best upgrade — your assistant can now look up exact request shapes, status codes, and webhook payloads on demand.
Bookmark these Markdown URLs and feed them to your agent at the start of a session:
* `https://docs.meld.io/docs/stablecoins/crypto-overview_/index.md` — product overview and which integration to pick
* `https://docs.meld.io/docs/stablecoins/crypto-overview_/quickstart.md` — first API call
* `https://docs.meld.io/docs/stablecoins/for-all-products/webhook-events.md` — webhook event catalog
When you want a clean-slate model (no MCP) to do something self-contained — a one-off script, a code review, a migration — paste `llms-full.txt` into context with caching enabled.
The MCP filesystem includes the OpenAPI JSON for every product. Ask your agent to validate generated requests against `/openapi/crypto-20260203.json` before you ship.
## Questions or feedback
Found a docs gap your agent stumbled on, or want a new endpoint covered? Reach out to your Meld contact — agent-readability is a first-class goal for this section of the docs.