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

# Changelog

> What changed and when across WalletSuite docs and products.

## 2026-07-23 - Hosted agent platform

**Give every AI agent a financial identity.** Configure and embed WalletSuite agents through one API — WalletSuite operates the models, tools, and financial execution controls behind the scenes.

* New: each agent is a scoped, revocable principal with its own key, used server-side to authorize the agent. See [Connect your agent](/ai-agents/token-exchange).
* New: capability is granted per agent through its tool profile and token scopes across four [execution levels](/ai-agents/execution-levels) — read, prepare, execute, governed autonomous.
* New: five surfaces on one platform — admin assistant, embedded assistant, customer-owned agent, background adaptive agent, deterministic workflow. See [Choose your surface](/ai-agents/choose-your-setup).

## 2026-07-13 - Organization policy engine

**A deterministic policy decision point now evaluates every signing intent before it can be signed.** No LLM sits in the decision path. A denied intent is never signed.

* New: deny-by-default org allowlist rules that match on chain, operation type, contract address, method selector, an optional destination pin, and an optional per-transaction amount cap in base units. Every rule carries a mandatory expiry.
* New: intents are bounded by the chains you're entitled to; operations that can't be typed are denied once the allowlist is armed.
* New: a decision record on every intent, a `policyDecisionId` plus the decision reason and the matched rule, written to the signing-intents ledger for allowed and denied intents alike. See [Policy Decisions](/core-concepts/policy-decisions).
* New: admin API `/v1/admin/allowlist-rules` to create, list, and deactivate rules, with a full audit snapshot on every change.
* New: rules roll out in stages: off, shadow, enforce.

## 2026-04-26 - MPC threshold signing

**Hosted MPC threshold signing is the signing model at all paid tiers.** WalletSuite holds one key share, the customer holds the other; both signatures are required for every transaction. No full private key ever exists on either side.

* New: hosted MPC co-signer (envelope-encrypted at rest behind the scenes, M-of-N internal access)
* Updated: `sign_transaction` and `send_transaction` route through MPC
* Updated: agent tokens carry policy bindings (chain allowlist, expiry, value caps) bound at issuance

## 2026-04-16 — Documentation Update

**Use Case Guides** added:

* [Give an AI Agent a Wallet](/getting-started/use-case-guides/give-ai-agent-a-wallet) — end-to-end guide from connect to first query in under 5 minutes

**Resources:**

* [Production Checklist](/production-checklist) — pre-go-live validation across API keys, policies, error handling, monitoring, and security

## 2026-04-01 — MCP Server v1.0

**Initial public release:**

* 7 read tools: `get_balance`, `get_all_balances`, `get_price`, `get_fee_quote`, `resolve_asset`, `get_tx_status`, `get_tx_history`
* 3 prepare tools: `prepare_transfer`, `prepare_serialized_unsigned_tx`, `prepare_onramp`
* Signing tools: `get_wallet_address`, `sign_transaction`
* 1 broadcast tool: `send_transaction`
* Policy gates on signing tools
* 5 install guides: LangChain, CrewAI, generic MCP runtime (Pydantic AI, Claude Agent SDK), Claude Desktop, Claude Code
* Non-custodial architecture: WalletSuite holds one key share, the customer holds the other, and no full private key ever exists on either side
* Structured error taxonomy with 6 categories and programmatic recovery
* Multi-chain support for read/prepare operations