# RadixScan Agent Layer

> The clearest, safest way for AI agents to read, understand, and build transactions on the Radix network. Radix transaction manifests are human- and machine-readable, statically analyzable, and free to simulate — so an agent can explain exactly what a transaction will do before a human signs it.

Status: live on mainnet and stokenet. The keyless Read+Build MCP server is deployed at `mcp.ai.radixscan.io`, with non-custodial, human-in-the-loop signing.

## Discovery

- [Landing page](https://ai.radixscan.io/): what this is and how to integrate
- [llms-full.txt](https://ai.radixscan.io/llms-full.txt): detailed, machine-oriented tool catalogue
- [/.well-known/agent.json](https://ai.radixscan.io/.well-known/agent.json): machine-readable endpoint and capability pointer
- [Explorer deep links](https://dashboard.radixscan.io/llms.txt): human-facing detail pages (`/transaction/{hash}`, `/account/{address}`, …) for sharing results with a human

## Interface

MCP (Model Context Protocol) over Streamable HTTP at `https://mcp.ai.radixscan.io/mcp` (SSE transport at `https://mcp.ai.radixscan.io/sse`). Keyless for read & build; signing is non-custodial and human-in-the-loop (Radix Wallet) — the server never holds keys.

- Read tools: radix_resolve_address, radix_get_account, radix_get_resource, radix_get_transaction, radix_list_transactions (recent transactions with dashboard-style filters), radix_get_validator, radix_get_validator_list, radix_get_blueprint_interface (the callable interface — "ABI" — of any package or component), radix_get_component_state (the stored, SBOR-decoded state of any component), radix_get_key_value_store (the entries behind a component's KeyValueStore references), radix_get_nft (NFT collection ids, per-NFT data and current holder), radix_get_resource_holders (the top holders of any resource — the reverse of radix_get_account), radix_find_entities_by_badge (the on-ledger entities a badge controls — the reverse of an entity's role assignments)
- Build tools (keyless): radix_build_transfer, radix_build_manifest, radix_validate_manifest, radix_preview_transaction, radix_explain_manifest
- Manifest documentation tools: radix_manifest_workflow, radix_manifest_instructions, radix_manifest_instruction, radix_manifest_templates, radix_manifest_template
- Ecosystem discovery: radix_list_ecosystem_services, radix_lookup_entity (curated name/symbol → canonical address for well-known tokens and dApp components)
- Signing tools (non-custodial, human-in-the-loop): radix_request_signature, radix_signature_status
- Resources: radix://manifest/workflow, radix://manifest/instructions, radix://manifest/instruction/{name}, radix://manifest/templates, radix://manifest/template/{name}, radix://ecosystem/services, radix://ecosystem/service/{slug}, radix://ecosystem/addresses/{network}, radix://blueprint/{address}

## Networks

- mainnet and stokenet — every tool is explicitly network-scoped

## Companion (build-time knowledge)

This server is runtime infrastructure: it reads the live ledger and builds, validates and previews transaction manifests over MCP. If your agent is also *writing* Radix software — Scrypto blueprints, dApps, radix-web3.js or TypeScript tooling — pair it with a build-time knowledge skill:

- [radix-skills](https://github.com/xstelea/radix-skills) — a community Agent Skill with reference guides for Scrypto, manifests, Gateway and wallet/ROLA flows, grounded in the radixdlt-scrypto and radix-web3.js source. Install: `npx skills add xstelea/radix-skills --skill radix`

The two are complementary, not overlapping: live data + transaction building over the wire here; local coding knowledge on disk there. radix-skills is maintained independently, not by RadixScan.

## Principles

- Keyless read & build; signing is human-in-the-loop and non-custodial (Radix Wallet)
- Free transaction preview before any recommendation
- No personal data and no user tracking; only anonymous, aggregate operational metrics are recorded (no IP, user, or address)

## Source

- [RadixScan](https://radixscan.io)