Namera Brings Permissioned Wallets to AI Agents

Overview

A new layer of financial infrastructure designed specifically for artificial intelligence agents has emerged, addressing one of the most persistent challenges in decentralized autonomous systems: secure, scoped transactional authority. The platform, known as Namera, introduces a model where AI agents can transact on blockchain networks without requiring unrestricted access to user funds.

This innovation shifts the paradigm from traditional account management to permissioned wallets, allowing developers and enterprises to grant AI tools the ability to spend, swap, and interact with smart contracts while maintaining strict human oversight. By embedding permissions directly into the session keys that agents use, Namera aims to reconcile the need for autonomous action with the non-negotiable requirement of security.

"Give agents the power to transact. You set what they can spend and do."

The solution targets a growing ecosystem of agentic commerce, trading desks, and on-chain games where automated entities must handle assets reliably. As AI models become more capable of executing multi-step workflows, the risk of a compromised or hallucinating agent draining a wallet has become a critical barrier to adoption. Namera’s approach ensures that an agent’s capabilities are bounded by policy, not just by the cryptographic key it holds.

How Permissioned Wallets Work

At the core of the Namera architecture is the concept of the session key. Unlike traditional private keys that provide blanket access to all associated funds, these keys are dynamically generated with a predefined scope. When an AI agent, such as one powered by Claude or Codex, initiates a transaction, the system intercepts the request and validates it against a set of rules established by the human owner.

Scoped Permissions and Limits

Every permissioned wallet operates within a rigid framework defined at issuance. Users can specify exact constraints, including:

Spending caps: Limits on the amount of specific tokens that can be spent per transaction or over a set period.
Allowed networks: Restricting the agent to specific blockchains, such as Ethereum or Base, while blocking access to all others.
Contract whitelisting: Ensuring the agent can only interact with verified protocols, such as Uniswap, preventing interactions with malicious contracts.
Expiration dates: Automatically revoking access after a set time, reducing the window of vulnerability.

If an agent attempts to exceed its allowance, the transaction is not merely flagged—it is blocked before it reaches the blockchain. For instance, if a default swap of 24 USDC is permitted but an agent attempts to swap 250 USDC, the policy engine rejects the operation immediately. This checked-before-signed mechanism ensures that the agent never possesses the authority to alter its own budget or expand its reach.

Key Custody and Security

A common concern in the crypto space is the custody of private keys. Namera emphasizes a zero-knowledge custody model where the platform never holds signing material. The session keys remain on the user’s machine, managed locally by the Namera binary. This architecture means that even if the Namera servers were compromised, the attacker would have no access to the actual signing keys required to move funds.

Furthermore, the account itself is owned by a user’s passkey, providing a familiar and secure entry point for identity verification. When multiple agents are deployed, each operates under its own isolated session key. This isolation means that a compromise in one agent’s environment does not jeopardize the permissions or assets associated with another.

Integration with AI Agents

The platform is designed to integrate seamlessly with existing Model Context Protocol (MCP) clients, which are becoming the standard for connecting AI assistants to external tools and data sources. Rather than requiring complex API integrations or exposing sensitive credentials, Namera acts as a secure middleware layer between the AI and the blockchain.

Supported Client Ecosystem

Developers can connect Namera to popular AI coding and automation tools with minimal configuration. The integration model supports:

Claude Code: Users can register the Namera MCP server, allowing the coding assistant to execute financial transactions within the bounds of its assigned session key.
Codex: Similar to Claude, OpenAI’s Codex can be pointed at the Namera binary to enable autonomous trading or transaction execution.
Any MCP Client: Because Namera runs as a local binary, any client that supports the MCP standard can connect without opening network ports or pasting secrets into configuration files.

This flexibility allows for a many-to-one or one-to-many relationship between users and agents. A single user might operate a trading agent with a $1,000 daily limit and a separate steward agent for a game with a $50 weekly budget. Each agent is treated as a distinct entity with its own credentials, and permissions can be revoked independently without affecting the others.

Transaction Lifecycle

The journey of an AI-initiated transaction follows a rigorous pipeline designed for transparency and safety. First, the intent is captured by the agent, specifying the action, amount, and target chain. The system then enters a prepare phase, where it performs a dry run of the transaction to estimate gas costs and verify feasibility. Notably, Namera can sponsor gas fees, removing the friction of users needing to hold native tokens for transaction costs.

Once prepared, the request is evaluated against the policy rules. If the spend cap, network, and contract all align with the permissions, the system proceeds to signing. The signature is generated locally on the user’s machine using the session key. Finally, the signed payload is submitted to the network, resulting in a confirmed execution on-chain. Throughout this process, the user retains full visibility through an activity dashboard that logs every signature, execution, and blocked attempt.

Industry Implications and Future Outlook

The introduction of permissioned wallets for AI agents signals a maturation in the way digital assets are managed. It addresses a specific pain point in the fintech and on-chain gaming industries, where trustless automation is essential but security cannot be compromised. By decoupling the ability to act from the ability to authorize broadly, Namera creates a safe sandbox for AI-driven economies.

Current dashboards show significant activity, with hundreds of executions and thousands of signatures logged over recent weeks. Early adopters include trading desks looking to automate portfolio rebalancing and agentic commerce platforms enabling bots to purchase goods and services autonomously.

As the regulatory landscape around AI and crypto continues to evolve, this model of auditable autonomy may become a standard requirement. The ability to prove that an agent acted strictly within predefined bounds provides a level of accountability that is crucial for institutional adoption. With the platform preparing for a broader launch, the focus remains on refining the user experience and expanding support for additional blockchain networks.

The underlying philosophy is clear: AI should be empowered to act, but humans must remain the ultimate gatekeepers of value. By ensuring that every agent operates within limits, Namera is laying the groundwork for a safer, more scalable future for autonomous economic agents.