# AI.tatum.io

> Machine-readable Tatum layer for AI agents: MCP, Skills.md, workflows, and JSON APIs.

Snapshot: 2026-06-04T07:42:00.276Z
Counts: 68 chains, 140 networks, 13 MCP tools, 2 OpenAPI schemas, 126 operations

This site is a machine-readable operating manual. Each HTML page has a Markdown twin at `<path>.md` with identical content, and every list is also exposed as JSON under `/api/*`. The single entry point for agents is `/.well-known/tatum-agent.json`.

## Agent routing

Workflows are end-to-end procedures for user goals. Capabilities map product areas to tools and APIs. MCP and Integration Skill are how you execute.

- If The user gave a concrete task or goal (monitor payments, check wallet, diagnose RPC) → Fetch /api/workflows first, then the matching /workflows/{slug}.md
- If You need to know what Tatum offers or which tools apply to a domain → Fetch /api/capabilities or browse /workflows#capabilities. Each entry links MCP tools, OpenAPI, docs, and embedded workflows
- If You will call Tatum at runtime inside an MCP client → Install MCP, then /api/mcp-tools
- If You will generate integration code into a repository → Use the Integration Skill (/integration-skill) and /auth for the API key
- If You need a specific RPC URL or to confirm Tatum supports a chain or network → Use /api/chains. One row per chain x network x interface, includes URLs. For the full nested tree use /api/blockchain-registry. For HTTP operation shapes use /api/openapi-index

## Agent manifest

- [Agent manifest (JSON)](https://ai.tatum.io/.well-known/tatum-agent.json): Entry point with counts, entrypoints, routing rules, endpoints, and example prompts.

## Entrypoints into Tatum

- [Blockchain MCP](https://ai.tatum.io/mcp.md): Runtime tool access for any MCP client. 13 tools across 130+ networks.
- [Skills.md](https://ai.tatum.io/integration-skill.md): Code-generation layer for coding agents (Cursor, Claude Code, OpenHands).

## Workflows

- [Workflows](https://ai.tatum.io/workflows.md): Procedures and capabilities in one index. Start here for user goals. JSON: `/api/workflows` and `/api/capabilities`.

## Reference

- [Blockchain Registry](https://ai.tatum.io/blockchains.md): Normalized registry of every supported chain, network, and RPC endpoint.
- [Open API](https://ai.tatum.io/openapi.md): Mirrored Blockchain Data and Notifications OpenAPI schemas.
- [Authentication](https://ai.tatum.io/auth.md): API-key placement.
- [Safety](https://ai.tatum.io/safety.md): Read-only defaults, address validation, human-in-the-loop for mutations.

## Machine-readable endpoints

- [Agent manifest (start here)](https://ai.tatum.io/.well-known/tatum-agent.json): `/.well-known/tatum-agent.json` (manifest)
- [Workflows](https://ai.tatum.io/api/workflows): `/api/workflows` (json)
- [Capabilities (+ linked workflows)](https://ai.tatum.io/api/capabilities): `/api/capabilities` (json)
- [MCP tools](https://ai.tatum.io/api/mcp-tools): `/api/mcp-tools` (json)
- [Chain endpoints (flat, one row per interface)](https://ai.tatum.io/api/chains): `/api/chains` (json)
- [Blockchain registry (nested tree)](https://ai.tatum.io/api/blockchain-registry): `/api/blockchain-registry` (json)
- [OpenAPI index](https://ai.tatum.io/api/openapi-index): `/api/openapi-index` (json)
- [LLM-friendly URL index (llmstxt.org convention)](https://ai.tatum.io/llms.txt): `/llms.txt` (text)
- [Every page concatenated as one Markdown document](https://ai.tatum.io/llms-full.txt): `/llms-full.txt` (text)

## Upstream

- [Tatum docs](https://docs.tatum.io): Canonical Tatum API reference.
- [Dashboard](https://dashboard.tatum.io): API key signup.
- [blockchain-mcp repo](https://github.com/tatumio/blockchain-mcp)
- [tatum-integration-skill repo](https://github.com/tatumio/tatum-integration-skill)

## Optional

- [Full content (concatenated)](https://ai.tatum.io/llms-full.txt): Same index, but with every page body inlined. Use this if your client can't follow links.