> For the complete documentation index, see [llms.txt](https://docs.eesel.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.eesel.ai/apis-and-developer-resources/allowing-eesel-through-your-firewall.md).

# Allowing eesel through your firewall

Which address eesel's requests arrive from, which of them carry our user agent, and which reach you from a provider instead, so you can let your agent through Cloudflare, a WAF or an IP allowlist.

If your store, help center, website or API sits behind Cloudflare, a WAF or an IP allowlist, our requests can be refused before they reach you. It usually looks like an integration that connects fine but reads nothing, product cards that never load, or a website crawl that finds no pages.

What you need to allow depends on which part of eesel is calling you.

| What is calling you                                                                                                                                        | Where it comes from                                                                                                                                                                                     |
| ---------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| An integration eesel reads or writes itself, like the storefront read behind your chat bubble's product cards, or a self-hosted helpdesk or knowledge site | eesel's servers, `50.19.245.229`                                                                                                                                                                        |
| A website crawl for knowledge                                                                                                                              | our crawling provider's cloud, not our address                                                                                                                                                          |
| Actions on an integration that connects through Pipedream, our connection provider, like WooCommerce order and refund actions                              | Pipedream's addresses, without our user agent                                                                                                                                                           |
| An API you add under [Your APIs](https://docs.eesel.ai/apis-and-developer-resources/network-access.md)                                                     | eesel's servers, `50.19.245.229`, if you saved an authentication header for it on a current workspace; otherwise your agent's own sandbox, which has no fixed address. See the Your APIs question below |

If you are locking one integration down and want to be sure, tell us which one and we will confirm the path it takes.

## eesel's own address

Everything eesel's own servers send you leaves from a single address:

```
50.19.245.229
```

It is one address, not a range, so add it as `50.19.245.229/32`. It is the only address our servers use, and it has not changed. If your security team needs to be told before it ever does, let us know.

It is not everything that can reach you, though. The crawl and Pipedream rows above come from those providers, so an IP rule built on our address alone will still refuse them.

## eesel's user agent

Where we identify ourselves, our requests carry this user agent:

```
eesel/1.0 (+https://www.eesel.ai)
```

The storefront read your chat bubble's product cards depend on is one of those, which makes it a good rule to pair with the IP. Not every request carries it: several of our helpdesk and knowledge reads go out as a plain HTTP client, so allow us by address as well, not instead.

{% hint style="info" %}
**Still refused after allowing us?** Check your bot protection as well as your firewall. They are separate rules in Cloudflare and in most WAFs, and allowing an address or a user agent in one does not always exempt it from the other.
{% endhint %}

**If the same read gets through one minute and is refused the next, it is not a path or an address rule.** Those either allow a request every time or refuse it every time. Something that lets a request through one second and turns the identical one away the next is sampling your traffic, which means bot protection or a rate limit rather than the firewall rule you have been editing. Widening the path again will not change it. Look at your bot protection and at any rate limit covering the API's path, and if your WAF logs a ray id or rule id for the refusals, that is the thing that names which of the two acted.

## If neither the address nor the user agent fits

Actions on an integration that connects through Pipedream reach you from Pipedream's addresses and do not carry our user agent, so neither rule above covers them. On a self-hosted store or helpdesk that is most of what your agent does: looking an order up, searching for a customer, issuing a refund.

Allow those by the API's path instead, and allow the whole prefix rather than the paths you tested. Your agent's actions reach many paths under one API and you will only ever have tried some of them. On WooCommerce that means everything under `/wp-json/wc/`, not `/wp-json/wc/v3/orders` on its own: listing orders and opening one order by its number are different paths, so a rule naming the first still refuses the second, and your agent ends up able to see that an order exists but not to read it.

If you would rather identify the requests than open the path, [get in touch](https://docs.eesel.ai/support/contact-us.md) and tell us which integration you are locking down, and we will tell you what your rule can match on.

## If a website crawl is being blocked

Website knowledge is fetched by our crawling provider rather than from our own servers, so neither our address nor our user agent will get a crawl through. Two ways round it:

* **Allow the crawl with a header.** The website connect form takes custom headers, so you can set a header that only you and eesel know and have your WAF allow a request carrying it. See [Website knowledge](https://docs.eesel.ai/integrations/knowledge/website.md).
* **Ask us.** If your WAF can only match on IP, get in touch and we will find out what your rule needs for crawl traffic.

## FAQ

<details>

<summary>Does eesel have a static IP address?</summary>

Yes, for our own servers: their requests arrive from `50.19.245.229`. Add it to an allowlist as `50.19.245.229/32`.

</details>

<details>

<summary>Can I have your full IP range?</summary>

`50.19.245.229/32` is the whole of it for traffic from eesel's servers, and there is no wider range to send. It is not everything that reaches you, though: website crawls come from our crawling provider, and actions on a Pipedream-connected integration come from Pipedream. Tell us which integration you are locking down and we will confirm which of those applies.

</details>

<details>

<summary>Can I allow eesel by user agent instead of by IP?</summary>

Not on its own. `eesel/1.0 (+https://www.eesel.ai)` covers the requests where we identify ourselves, and several of our helpdesk and knowledge reads do not, so pair it with the address rather than replacing it.

</details>

<details>

<summary>My store connects, but the agent can't read anything from it. Is my firewall the cause?</summary>

Often, yes. Your store's API answering the open internet is not the same as it answering us. Look in your WAF or bot protection logs for our user agent or our address around the time the agent tried, and if you find us being challenged or blocked, allow us by either.

</details>

<details>

<summary>The same read works sometimes and is refused other times.</summary>

Then a path rule is not what is refusing it, because those are all-or-nothing. Intermittent refusals come from bot protection or a rate limit, which are separate settings from your firewall rules. Check both, and look up one of the refusals in your WAF's own event log to see which acted.

</details>

<details>

<summary>Some reads always work and others always fail. Listing orders works every time, opening one never does.</summary>

Usually a path rule at your edge naming the paths someone tested rather than the whole API. Listing orders and opening a single order are different paths under the same API, so a rule built on the first leaves the second refused. Widen it to the whole prefix, which on WooCommerce is everything under `/wp-json/wc/`.

</details>

<details>

<summary>Should I restrict an API I added under Your APIs to your address?</summary>

It depends on what you saved for it.

* **You saved an authentication header.** On current workspaces your agent's calls to that API go out from our servers, so `50.19.245.229/32` is the rule you need. On some older workspaces they still go out from the agent's own sandbox.
* **You allowed the host with no header.** Your agent calls it from its own sandbox, which has no fixed address, so an IP rule would refuse your agent.

Tell us the host and we will confirm which applies to you before you lock it down.

Either way, authenticating the API with a header is what protects it; the address only narrows who can reach it. If you allowed a host with no header, adding one is the safer change.

</details>

## Related pages

* [Your APIs](https://docs.eesel.ai/apis-and-developer-resources/network-access.md) for letting your agent call an API of your own
* [Website knowledge](https://docs.eesel.ai/integrations/knowledge/website.md) for what a crawl reads, and the headers you can set on it
* [Security and Privacy](https://docs.eesel.ai/account/security-and-privacy.md) for how your data is handled
* [Contact us](https://docs.eesel.ai/support/contact-us.md) if you are still blocked after allowing us


---

# 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.eesel.ai/apis-and-developer-resources/allowing-eesel-through-your-firewall.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.
