AgentArm - Policy Configuration

AgentArm

► Policy Configuration

Deploy

Agents

Logs

Config

Security Policies

⚠ Phase 2 Feature Preview — Available in AgentArm Business ($50/100K events)

✓

Block Dangerous Commands

BLOCK

Prevent execution of dangerous system commands

Rule: tools matching /rm -rf|DROP TABLE|DELETE FROM/

⚠

Require Approval for Deployments

REQUIRE_APPROVAL

All deployment operations need human approval

Rule: tools matching /deploy|release|publish/

✓

Mask PII Data

MASK

Automatically detect and mask personally identifiable information

Patterns: SSN, credit cards, email addresses, phone numbers

⚠

Rate Limit LLM Calls

RATE_LIMIT

Prevent cost overruns from runaway agents

Limit: 100 LLM calls per hour per agent

○

Block External API Calls

BLOCK

Prevent agents from calling unapproved external APIs

Allowlist: github.com, slack.com, internal.acme.com