Skip to content
Arxo Arxo

Policy Examples

Policy Examples

Section titled “Policy Examples”

This guide shows policy examples for the Agent Architecture metric. Invariants use metric keys from that metric.

Understanding Policies

Section titled “Understanding Policies”

Policies in Arxo are defined in YAML and consist of invariants — conditions that must hold for the run to pass.

policy:
  invariants:
    - metric: <metric_id.key>
      op: <operator>
      value: <threshold>
      message: <optional_message>

Operators: ==, !=, <, <=, >, >=

Metric keys are scoped (e.g. agent_architecture.overall_agent_health). See Scoring and Keys for the full list.

Section titled “Agent Architecture Gate (Recommended)”

Use Case: CI and local checks for agent reliability and governance

metrics:
  - id: agent_architecture


policy:
  invariants:
    - metric: agent_architecture.loop_guard_absence
      op: "<="
      value: 0.20
      message: "Agent loops must be bounded with step/time guards"
    - metric: agent_architecture.governance_readiness
      op: ">="
      value: 80
      message: "Tool governance baseline not met"
    - metric: agent_architecture.overall_agent_health
      op: ">="
      value: 0.70
      message: "Overall agent health below minimum threshold"

Run:

Terminal window
arxo analyze --fail-fast
# or with config
arxo analyze --config .arxo.yaml

See Agent Architecture policy and CI gates for strict/pragmatic/no-regression profiles.

Safety and Coordination

Section titled “Safety and Coordination”

Focus on safety and coordination axes:

metrics:
  - id: agent_architecture


policy:
  invariants:
    - metric: agent_architecture.safety_protocol_score
      op: ">="
      value: 75
      message: "Safety protocol score too low"
    - metric: agent_architecture.coordination_maturity_score
      op: ">="
      value: 70
      message: "Coordination maturity below threshold"

Environment-Specific Policies

Section titled “Environment-Specific Policies”

Use different configs per environment:

Development (.arxo.dev.yaml):

metrics:
  - id: agent_architecture


policy:
  invariants:
    - metric: agent_architecture.overall_agent_health
      op: ">="
      value: 0.5
      message: "Relaxed threshold for local dev"

CI (.arxo.ci.yaml):

metrics:
  - id: agent_architecture


policy:
  invariants:
    - metric: agent_architecture.overall_agent_health
      op: ">="
      value: 0.70
      message: "Overall agent health below minimum"

Usage:

Terminal window
arxo analyze --config .arxo.dev.yaml
arxo analyze --config .arxo.ci.yaml

Custom Messages

Section titled “Custom Messages”

Provide actionable guidance:

policy:
  invariants:
    - metric: agent_architecture.overall_agent_health
      op: ">="
      value: 0.70
      message: |
        Agent architecture health is below 0.70.


        Check:
        1. Run: arxo analyze --format json | jq '.results[] | select(.id=="agent_architecture")'
        2. Review axis scores (governance, reliability, safety, coordination)
        3. See: /metrics/agent-architecture/remediation-playbook

MCP

Section titled “MCP”

Use the MCP server to run analysis and get results; policy is enforced by the engine based on your config:

You: "Run architecture analysis and check policy"
AI: [Uses analyze_architecture] "Analysis complete. No violations."

See MCP Workflows and analyze_architecture.

Next Steps

Section titled “Next Steps”