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

# Changelog

> What's new in AMEO.

<Update label="June 12, 2026" tags={["release"]}>
  ## New

  **Fresh AgentIdentity deployment (bytecode-verified)**
  Canonical contract `0xB86dC64573089D8DD89C5686010295bB4412D652` on Mantle Sepolia — Sourcify `exact_match` with local `forge build` artifact. Agent NFT token `0` minted to burner `0x59ffc8907beaA275F29B466BCB1D9BbfeaDAd165`.

  **`/v1/identity/status`**
  Preflight endpoint reports `ready: true` when the signing EOA owns the configured token id.

  ## Improvements

  * Policy caps (`250` / `500`) exposed consistently from `/v1/config` and `/health`.
  * DEX execution degrades to labeled `treasury_ping` on Sepolia revert instead of raw `execution reverted`.
  * LLM primary provider default: Groq.
</Update>

<Update label="May 29, 2026" tags={["release"]}>
  ## New

  **Public REST API (`/v1`)**
  A versioned JSON API for agents, decisions, verification, policies, and skills is now live at `https://agentic-micro-economy-os.onrender.com`. Build integrations against a stable surface without running the worker yourself. See the [REST API v1 reference](/api/v1).

  **TypeScript SDK — `@ameo/sdk`**
  An official SDK wraps the `/v1` API with a typed client for agent registration, decision creation, proof verification, and policy reads. Install with `npm install @ameo/sdk` and get started with the [SDK guide](/sdk).

  **MCP server — `@ameo/mcp`**
  Connect AMEO to Claude Desktop, Cursor, and other MCP hosts. The server exposes verify, decisions, and agent tools so you can audit on-chain decisions from your editor or chat. See the [MCP setup guide](/mcp).

  **`/v1/skills` endpoint**
  The execution-skill registry is now queryable over HTTP, MCP, and SDK. Discover which execution skills (for example `mantle.swap.v1`) an agent can invoke. See the [execution and quote telemetry integration guide](/guides/execution-quote-telemetry-integration).

  ## Improvements

  **Quote telemetry visibility**
  Every cycle now emits a structured quote-telemetry log line before signing and surfaces it in the replay view as the execution planning step. Operators can audit *what* ran without reading worker internals.

  **Render reliability**
  Background keepalive keeps the hosted worker warm, reducing cold-start delays on the public endpoint.

  **Canonical environment alignment**
  The dashboard, worker, and SDK now read a consistent runtime configuration, removing drift between hosted and local deployments. See [Run locally](/guides/run-locally).

  ## Fixes

  * Operator settings sheet and cognition timeline render correctly across viewport sizes.
  * Dashboard footer and proof links resolve against the active network on first load.
  * Mantlescan verification badge is now bundled and displays on verified proofs.
</Update>