> For the complete documentation index, see [llms.txt](https://docs.vultisig.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.vultisig.com/developer-docs/ai-agents/authority-and-security.md).

# Agent signing authority in Vultisig

Choose between autonomous, policy-bound, and human-approved agent signing models with Vultisig.

MPC removes the complete private key as a single point of failure. It does not make an agent's decisions correct. Choose a signing model based on the value at risk, the workflow's speed requirements, and the controls that must remain outside the agent.

## Compare signing models

### Autonomous Fast Vault

A Fast Vault combines an agent-controlled share with a VultiServer share. The agent can sign without a person present.

**Use when:** the workflow requires unattended execution and the vault can be funded according to a defined risk budget.

**Primary controls:** isolated vaults, limited funding, protected credentials, transaction previews, monitoring, and tested backups.

### Policy-bound Marketplace automation

{% hint style="warning" %}
**Paused.** The plugin marketplace is not in production. This signing model is not available to users or developers today.
{% endhint %}

A Marketplace plugin proposes unsigned transactions. The Verifier checks each proposal against the user's configured policy before participating in MPC signing.

**Use when:** an autonomous strategy should operate inside explicit limits, such as allowed assets, destinations, amounts, or timing.

**Primary controls:** narrowly scoped rules, Verifier policy validation, plugin review, monitoring, and the ability to disable the automation.

### Human-approved Secure Vault

A Secure Vault requires the configured threshold of devices to join each signing session. In a human-approved setup, the agent prepares the action and the human-controlled shares complete the signing threshold.

**Use when:** every transaction requires human review or multiple participants must approve.

**Primary controls:** transaction review, device separation, threshold configuration, backups, and clear approval procedures.

## Responsibilities that remain with the developer

Your application remains responsible for:

* Prompt and tool-call security
* Strategy logic and economic outcomes
* Token, contract, route, and destination selection
* Transaction parameters and fee limits
* Vault passwords, backups, logs, and runtime credentials
* Monitoring, incident response, and disabling compromised agents

## Recommended controls

* Start with read-only portfolio access.
* Require previews before enabling signing.
* Use separate vaults for separate agents or strategies.
* Fund autonomous vaults only to an explicit risk budget.
* Prefer policy-bound automation when rules can constrain the workflow.
* Prefer Secure Vaults for high-value or irreversible actions.
* Treat arbitrary message and byte signing as high-risk capabilities.
* Never embed vault passwords or backups in prompts, source code, or logs.

{% hint style="warning" %}
Fast Vaults enable unattended signing. Only grant an agent the assets and capabilities its workflow requires.
{% endhint %}


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.vultisig.com/developer-docs/ai-agents/authority-and-security.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.