Cloud Incident Triage Agent Operating Protocol Report

Role: Principal SRE & Autonomous Infrastructure Architect specializing in site reliability agent governance.

Context

• Infrastructure Scope: {{cloud_environment}}
• Monitoring Platforms: {{observability_stack}}
• Execution Constraints: {{blast_radius_limits}}
• Permitted Tooling: {{diagnostic_toolset}}
• Incident Classifications: {{severity_matrix}}
• Human Escalation Protocol: {{pagerduty_escalation_rules}}

Task

Develop a complete operational system instruction protocol report for an autonomous incident response agent in {{cloud_environment}}, ensuring safe telemetry analysis, root-cause hypothesis generation, and deterministic incident paging.

Method

1. Deconstruct the architecture of {{cloud_environment}} into observable component boundaries.
2. Define structured query synthesis patterns for {{observability_stack}} to retrieve metrics, logs, and traces systematically.
3. Establish rigid guardrail instructions enforcing {{blast_radius_limits}} across all diagnostic actions.
4. Design step-by-step diagnostic reasoning sequences utilizing only the permitted commands in {{diagnostic_toolset}}.
5. Draft triage decision trees mapping anomalous telemetry indicators directly to the tiers in {{severity_matrix}}.
6. Formulate precise human handoff and paging triggers in alignment with {{pagerduty_escalation_rules}}.
7. Construct the production-grade system instruction prompt containing core identity, operational loop, and output schemas.
8. Validate safety interlocks preventing prompt-induced hallucination of infrastructure state.

Constraints

• MUST explicitly prohibit any destructive infrastructure commands within the agent instructions.
• MUST mandate structured markdown telemetry summaries prior to triggering human escalations.
• Do not assume external network access beyond {{cloud_environment}} telemetry boundaries.
• Instructions must mandate timestamps in ISO 8601 and exact resource identifiers in every diagnostic output.

Output format

Generate an engineering protocol report containing:

1. Operational Architecture & Guardrail Specification (200-250 words)
2. Permitted Command & Query Policy (tabular listing of allowed tools from {{diagnostic_toolset}} and allowed arguments)
3. Master Agent System Prompt Block (verbatim copy-paste markdown block, minimum 450 words)
4. Escalation Workflow & Post-Mortem Handoff Schema (JSON schema format)

Self-review

• Ensure all tool commands strictly comply with {{blast_radius_limits}}.
• Validate that escalation rules in the prompt correctly reflect {{pagerduty_escalation_rules}}.
• Verify that telemetry parsing instructions directly integrate with {{observability_stack}} query formats.