AppsAI Agent Firewall

Policy and operations

Custom rules and exceptions

Custom rules let you shape the AI Agent Firewall like a network firewall: ordered rules add an action and exceptions remove findings, with up to 10 conditions each, and every rule can be previewed against the last 7 days of real traffic before you save it.

Turn on the firewall
New rulepreview: last 7 daysSAMPLE

WHEN · TOOL_CALL

tool is Change bank details

country is not US

person has not verified

Then BLOCK and stop.

Reads as: “Block Change bank details from outside the US unless the person verified.” Would have blocked 3 requests last week.

The problem

Every business has its own lines: bank changes never from abroad, payroll only in office hours, a trusted partner API that shouldn’t trip new-site alerts. Built-in checks can’t know them.

How custom rules and exceptions works

STEP 1

Choose Rule or Exception

Rules add a finding with an action. Exceptions remove built-in findings that don’t matter in a situation.

STEP 2

Choose when it applies

Message, tool call, tool result, reply or caller voice, or leave empty for all.

STEP 3

Add conditions

Up to 10, matched all or any. The editor reads the rule back in plain words.

STEP 4

Preview

Preview against the last 7 days shows how many requests it would have changed, with examples.

STEP 5

Order and stop

Rules run top to bottom; a rule with Stop ends evaluation. Allow + Stop is an explicit allow. The strongest remaining action wins.

Every setting

Actions

Allow, Flag, Ask to verify, Require approval, Block.

Condition fields

Channel, agent, tool (and whether it needs verification or returns sensitive records), identity and verified status, country, IP, hour and weekday, voice score, caller ID result, unusual-request score, prompt-injection score, pressure tactics found, sensitive-record counts, destination site or email domain, message or reply text, tool inputs, tool calls and rows used.

Text patterns

Up to 200 characters.

Limits

Up to 100 rules per policy.

Where rules live

Company policy, an agent’s custom policy, and named API policies.

Who can do what

Admins who manage team members

Create, order, preview and delete rules.

What to know

  • Exceptions can’t skip safety checks: the duress word, verification lockouts, a secret about to leave, a check that couldn’t run and a tool’s own Verify before use are protected.
  • Rules on message text or tool inputs can’t be previewed against past traffic, because that text isn’t kept with alerts; the preview says which.
  • In Monitor, rules only flag. In Enforce, their actions apply.

More in Policy and operations

Monitor and Enforce →Monitor records what every check would do without changing what customers see; Enforce applies the actions. Each agent uses the company policy, its own custom policy, or none.Alerts and approvals →Every decision lands on a sortable alerts page with the score, cues and a link to the conversation; held tool calls get Approve and Reject, also from the admin’s inbox.Monitoring dashboard →One dashboard shows checks, catches, blocks, verifications and points for every agent and API policy, with trends, what was caught, top rules and top identities.Attack simulator →A dry-run red team: built-in prompt injection, pressure, passphrase guessing, word replay, anomaly, exfiltration and cloned-voice attacks against an agent, with a scorecard of what got through.All of Policy and operations →

FAQ

Custom rules and exceptions: questions

Can I write my own AI guardrail rules?

Yes. Write ordered rules and exceptions with up to 10 conditions on channel, identity, tool, location, time, scores and text, and preview them against the last 7 days.

Can an exception turn off the duress word?

No. Protected checks can never be removed by an exception; such an exception is rejected when you save.

Turn on custom rules and exceptions.

Ordered rules add an action and exceptions remove findings, using conditions on channel, tool, identity, voice score, country, time and text, previewed against the last 7 days. Start in Monitor and see what it would catch on your real traffic.

Turn on the firewall

Starts in Monitor · nothing changes for your customers until you enforce

100% customizable, code and all

Make Autessa AI Agent Firewall fully your own.

Use it out of the box, then change anything. Ask Autobot, Autessa’s AI builder, for a new field, screen, report, workflow or agent and it builds it into your copy of the app. Or open the code and edit it yourself. Nothing is locked: the app, its data and its code are yours.

Start free and make it yours