> For the complete documentation index, see [llms.txt](https://doc.threatreaper.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://doc.threatreaper.ai/user-guide/ai-agents-autonomous-security.md).

# 🤖 AI-Agents — Autonomous Security

**Menu:** PROTECT > AI-Agents > Autonomous Agent.

The autonomous engine makes allow, block, challenge, or escalate decisions for requests sent to `/api/autonomous/analyze`, using the project policy, per-guardrail actions, allowed/blocked topics, and a per-user trust score.

This page summarises:

* Total decisions and the allow / block / challenge / escalate split
* Average trust score, risk score, and latency
* Decision trend over recent days
* Risk distribution (low, medium, high, critical)
* Top triggered guardrails and their block rate
* Tickets raised for follow-up and their state
* A table of recent decisions

## 7.1 How to configure it

AI-Agents has no separate setup screen—it runs on top of a policy's **Autonomous Features** step.

{% stepper %}
{% step %}

## Edit or create a policy

In **Settings > Policies**, edit or create a policy.
{% endstep %}

{% step %}

## Enable autonomous decisions

Turn on **Autonomous Enabled**.
{% endstep %}

{% step %}

## Configure each guardrail

For each guardrail, set:

* **Action** — Block, Allow, Challenge, or Escalate
* **Confidence Threshold** — how certain the model must be before acting (default `0.7`)
* **Allowed Topics** / **Blocked Topics** — comma-separated lists always treated as safe or unsafe
  {% endstep %}

{% step %}

## Optionally configure verification and ticketing

Optionally, enable **Require Verification** and configure a **SOAR** integration (Jira, ServiceNow, or another connected ITSM tool) to automatically raise a ticket when a guardrail fires.
{% endstep %}
{% endstepper %}

## 7.2 Errors you might see and how to fix them

| **Message**                                            | **What it means**                                                                        | **What to do**                                                                                |
| ------------------------------------------------------ | ---------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------- |
| Dashboard shows an error banner with a refresh icon    | The decision data couldn't be loaded                                                     | Select the refresh icon to retry; if it persists, check your connection or contact support    |
| All metrics show zero                                  | No autonomous decisions have been recorded yet for the selected project/date range       | Confirm a policy has **Autonomous Enabled** and is receiving traffic, or widen the date range |
| A request that should auto-resolve stays at "Escalate" | The per-guardrail confidence threshold is set too low, or **Require Verification** is on | Review the guardrail's action and threshold in **Settings > Policies**                        |

## 7.3 What success looks like

**Total decisions:** 18,420\
**Avg trust score:** 0.82\
**Avg latency:** 124ms

| Decision  |  Count | Percentage |
| --------- | -----: | ---------: |
| Allow     | 16,110 |      87.4% |
| Block     |  1,540 |       8.4% |
| Challenge |    580 |       3.1% |
| Escalate  |    190 |       1.1% |

| Top triggered guardrails | Block rate |
| ------------------------ | ---------: |
| PII Detection            |        62% |
| Prompt Injection         |        41% |
| Toxicity                 |        18% |

Each row in the **recent decisions** table shows the user, the decision (`ALLOW`/`BLOCK`/`CHALLENGE`/`ESCALATE`), the risk score, the guardrail(s) that triggered, and a link to the matching **AI-Explain** record so you can see exactly why that decision was made.


---

# 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://doc.threatreaper.ai/user-guide/ai-agents-autonomous-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.