open-vault/website/content/docs/agent/index.mdx

---
layout: docs
page_title: Vault Agent
description: |-
  Vault Agent is a client-side daemon that can be used to perform some Vault
  functionality automatically.
---

# Vault Agent

A valid client token must accompany most requests to Vault. This
includes all API requests, as well as via the Vault CLI and other libraries.
Therefore, Vault clients must first authenticate with Vault to acquire a token.
Vault provides several different authentication methods to assist in
delivering this initial token.

![Client authentication](/img/diagram-vault-agent.png)

If the client can securely acquire the token, all subsequent requests (e.g., request
database credentials, read key/value secrets) are processed based on the
trust established by a successful authentication.

This means that client application must invoke the Vault API to authenticate
with Vault and manage the acquired token, in addition to invoking the API to
request secrets from Vault. This implies code changes to client applications
along with additional testing and maintenance of the application.

The following code example implements Vault API to authenticate with Vault
through [AppRole auth method](/docs/auth/approle#code-example), and then uses
the returned client token to read secrets at `kv-v2/data/creds`.

```go
package main

import (
    ...snip...
    vault "github.com/hashicorp/vault/api"
)

// Fetches a key-value secret (kv-v2) after authenticating via AppRole
func getSecretWithAppRole() (string, error) {
    config := vault.DefaultConfig()

    client := vault.NewClient(config)
    wrappingToken := ioutil.ReadFile("path/to/wrapping-token")
    unwrappedToken := client.Logical().Unwrap(strings.TrimSuffix(string(wrappingToken), "\n"))

    secretID := unwrappedToken.Data["secret_id"]
    roleID := os.Getenv("APPROLE_ROLE_ID")

    params := map[string]interface{}{
        "role_id":   roleID,
        "secret_id": secretID,
    }
    resp := client.Logical().Write("auth/approle/login", params)
    client.SetToken(resp.Auth.ClientToken)

    secret, err := client.Logical().Read("kv-v2/data/creds")
    if err != nil {
        return "", fmt.Errorf("unable to read secret: %w", err)
    }

    data := secret.Data["data"].(map[string]interface{})

    ...snip...
}
```

For some Vault deployments, making (and maintaining) these changes to
applications may not be a problem, and may actually be preferred. This may be
applied to scenarios where you have a small number of applications or you want
to keep strict, customized control over how each application interacts with
Vault. However, in other situations where you have a large number of
applications, as in large enterprises, you may not have the resources or expertise
to update and maintain the Vault integration code for every application. When
third party applications are being deployed by the application, it is prohibited
to add the Vault integration code.

Vault Agent aims to remove this initial hurdle to adopt Vault by providing a
more scalable and simpler way for applications to integrate with Vault.


## What is Vault Agent?

Vault Agent is a client daemon that provides the following features:

- [Auto-Auth][autoauth] - Automatically authenticate to Vault and manage the
token renewal process for locally-retrieved dynamic secrets.
- [Caching][caching] - Allows client-side caching of responses containing newly
created tokens and responses containing leased secrets generated off of these
newly created tokens. The agent also manages the renewals of the cached tokens and leases.
- [Windows Service][winsvc] - Allows running the Vault Agent as a Windows
service.
- [Templating][template] - Allows rendering of user-supplied templates by Vault
Agent, using the token generated by the Auto-Auth step.


## Auto-Auth

Vault Agent allows easy authentication to Vault in a wide variety of
environments. Please see the [Auto-Auth docs][autoauth]
for information.

Auto-Auth functionality takes place within an `auto_auth` configuration stanza.

## Caching

Vault Agent allows client-side caching of responses containing newly created tokens
and responses containing leased secrets generated off of these newly created tokens.
Please see the [Caching docs][caching] for information.

## API

### Quit

This endpoints triggers shutdown of the agent. By default, it is disabled, and can
be enabled per listener using the [`agent_api`][agent-api] stanza. It is recommended
to only enable this on trusted interfaces, as it does not require any authorization to use.

| Method | Path             |
| :----- | :--------------- |
| `POST` | `/agent/v1/quit` |

### Cache

See the [caching](/docs/agent/caching#api) page for details on the cache API.

## Configuration

These are the currently-available general configuration option:

- `vault` <code>([vault][vault]: <optional\>)</code> - Specifies the remote Vault server the Agent connects to.

- `auto_auth` <code>([auto_auth][autoauth]: <optional\>)</code> - Specifies the method and other options used for Auto-Auth functionality.

- `cache` <code>([cache][caching]: <optional\>)</code> - Specifies options used for Caching functionality.

- `listener` <code>([listener][listener]: <optional\>)</code> - Specifies the addresses and ports on which the Agent will respond to requests.

- `pid_file` `(string: "")` - Path to the file in which the agent's Process ID
  (PID) should be stored

- `exit_after_auth` `(bool: false)` - If set to `true`, the agent will exit
  with code `0` after a single successful auth, where success means that a
  token was retrieved and all sinks successfully wrote it

- `template` <code>([template][template]: <optional\>)</code> - Specifies options used for templating Vault secrets to files.

- `template_config` <code>([template_config][template-config]: <optional\>)</code> - Specifies templating engine behavior.

- `telemetry` <code>([telemetry][telemetry]: <optional\>)</code> – Specifies the telemetry
  reporting system. See the [telemetry Stanza](/docs/agent#telemetry-stanza) section below
  for a list of metrics specific to Agent.

### vault Stanza

There can at most be one top level `vault` block and it has the following
configuration entries:

- `address` `(string: <optional>)` - The address of the Vault server. This should
  be a complete URL such as `https://127.0.0.1:8200`. This value can be
  overridden by setting the `VAULT_ADDR` environment variable.

- `ca_cert` `(string: <optional>)` - Path on the local disk to a single PEM-encoded
  CA certificate to verify the Vault server's SSL certificate. This value can
  be overridden by setting the `VAULT_CACERT` environment variable.

- `ca_path` `(string: <optional>)` - Path on the local disk to a directory of
  PEM-encoded CA certificates to verify the Vault server's SSL certificate.
  This value can be overridden by setting the `VAULT_CAPATH` environment
  variable.

- `client_cert` `(string: <optional>)` - Path on the local disk to a single
  PEM-encoded CA certificate to use for TLS authentication to the Vault server.
  This value can be overridden by setting the `VAULT_CLIENT_CERT` environment
  variable.

- `client_key` `(string: <optional>)` - Path on the local disk to a single
  PEM-encoded private key matching the client certificate from `client_cert`.
  This value can be overridden by setting the `VAULT_CLIENT_KEY` environment
  variable.

- `tls_skip_verify` `(string: <optional>)` - Disable verification of TLS
  certificates. Using this option is highly discouraged as it decreases the
  security of data transmissions to and from the Vault server. This value can
  be overridden by setting the `VAULT_SKIP_VERIFY` environment variable.

- `tls_server_name` `(string: <optional>)` - Name to use as the SNI host when
  connecting via TLS. This value can be overridden by setting the
  `VAULT_TLS_SERVER_NAME` environment variable.

#### retry Stanza

The `vault` stanza may contain a `retry` stanza that controls how failing Vault
requests are handled, whether these requests are issued in order to render
templates, or are proxied requests coming from the proxy cache subsystem.
Auto-auth, however, has its own notion of retrying and is not affected by this
section.

For requests from the templating engine, Agent will reset its retry counter and
perform retries again once all retries are exhausted. This means that templating
will retry on failures indefinitely unless `exit_on_retry_failure` from the
[`template_config`][template-config] stanza is set to `true`.

Here are the options for the `retry` stanza:

- `num_retries` `(int: 12)` - Specify how many times a failing request will
  be retried. A value of `0` translates to the default, i.e. 12 retries.
  A value of `-1` disables retries. The environment variable `VAULT_MAX_RETRIES`
  overrides this setting.

There are a couple of subtleties to be aware of here. First, requests originating
from the proxy cache will only be retried if they resulted in specific HTTP
result codes: any 50x code except 501 ("not implemented"), as well as 412
("precondition failed"); 412 is used in Vault Enterprise 1.7+ to indicate a
stale read due to eventual consistency. Requests coming from the template
subsystem are retried regardless of the failure.

Second, templating retries may be performed by both the templating engine _and_
the cache proxy if Agent [persistent
cache][persistent-cache] is enabled. This is due to the
fact that templating requests go through the cache proxy when persistence is
enabled.

Third, the backoff algorithm used to set the time between retries differs for
the template and cache subsystems. This is a technical limitation we hope
to address in the future.

### listener Stanza

Agent supports one or more [listener][listener_main] stanzas. In addition to
the standard listener configuration, an Agent's listener configuration also
supports the following:

- `require_request_header` `(bool: false)` - Require that all incoming HTTP
  requests on this listener must have an `X-Vault-Request: true` header entry.
  Using this option offers an additional layer of protection from Server Side
  Request Forgery attacks. Requests on the listener that do not have the proper
  `X-Vault-Request` header will fail, with a HTTP response status code of `412: Precondition Failed`.

- `agent_api` <code>([agent_api][agent-api]: <optional\>)</code> - Manages optional Agent API endpoints.

#### agent_api Stanza

- `enable_quit` `(bool: false)` - If set to `true`, the agent will enable the [quit](/docs/agent#quit) API.

### telemetry Stanza

Vault Agent supports the [telemetry][telemetry] stanza and collects various
runtime metrics about its performance, the auto-auth and the cache status:

| Metric                           | Description                                          | Type    |
| -------------------------------- | ---------------------------------------------------- | ------- |
| `vault.agent.auth.failure`       | Number of authentication failures                    | counter |
| `vault.agent.auth.success`       | Number of authentication successes                   | counter |
| `vault.agent.proxy.success`      | Number of requests successfully proxied              | counter |
| `vault.agent.proxy.client_error` | Number of requests for which Vault returned an error | counter |
| `vault.agent.proxy.error`        | Number of requests the agent failed to proxy         | counter |
| `vault.agent.cache.hit`          | Number of cache hits                                 | counter |
| `vault.agent.cache.miss`         | Number of cache misses                               | counter |



## Start Vault Agent

To run Vault Agent:

1. [Download](/downloads) the Vault binary where the client application runs
(virtual machine, Kubernetes pod, etc.)

1. Create a Vault Agent configuration file. (See the [Example
Configuration](#example-configuration) section for an example configuration.)

1. Start a Vault Agent with the configuration file.

   **Example:**

   ```shell-session
   $ vault agent -config=/etc/vault/agent-config.hcl
   ```

   To get help, run:

   ```shell-session
   $ vault agent -h
   ```

## Example Configuration

An example configuration, with very contrived values, follows:

```python
pid_file = "./pidfile"

vault {
  address = "https://127.0.0.1:8200"
  retry {
    num_retries = 5
  }
}

auto_auth {
  method "aws" {
    mount_path = "auth/aws-subaccount"
    config = {
      type = "iam"
      role = "foobar"
    }
  }

  sink "file" {
    config = {
      path = "/tmp/file-foo"
    }
  }

  sink "file" {
    wrap_ttl = "5m"
    aad_env_var = "TEST_AAD_ENV"
    dh_type = "curve25519"
    dh_path = "/tmp/file-foo-dhpath2"
    config = {
      path = "/tmp/file-bar"
    }
  }
}

cache {
  use_auto_auth_token = true
}

listener "unix" {
  address = "/path/to/socket"
  tls_disable = true
}

listener "tcp" {
  address = "127.0.0.1:8100"
  tls_disable = true
}

template {
  source = "/etc/vault/server.key.ctmpl"
  destination = "/etc/vault/server.key"
}

template {
  source = "/etc/vault/server.crt.ctmpl"
  destination = "/etc/vault/server.crt"
}
```

[vault]: /docs/agent#vault-stanza
[autoauth]: /docs/agent/autoauth
[caching]: /docs/agent/caching
[persistent-cache]: /docs/agent/caching/persistent-caches
[template]: /docs/agent/template
[template-config]: /docs/agent/template-config
[agent-api]: /docs/agent/#agent_api-stanza
[listener]: /docs/agent#listener-stanza
[listener_main]: /docs/configuration/listener/tcp
[winsvc]: /docs/agent/winsvc
[telemetry]: /docs/configuration/telemetry