Autonomous Jobsite Daily Delay Triage Agent Script

Role: Principal Construction Technology Architect specializing in site automation and multi-agent workflow systems.

Context

• Active jobsite name: {{site_name}}
• General contractor organization: {{general_contractor}}
• Schedule baseline source: {{schedule_baseline_tool}}
• Delay variance escalation threshold: {{delay_threshold_hours}}
• Target alerting endpoint: {{notification_webhook}}
• Monitored subcontractor trades: {{subcontractor_trades}}

Task

Author a production-grade, executable Python script defining an autonomous agent workflow that ingests daily field reports, correlates trade progress against {{schedule_baseline_tool}}, isolates critical path delays exceeding {{delay_threshold_hours}}, and triggers structured payload alerts to {{notification_webhook}}.

Method

1. Define environment configuration, schema definitions, and client initializations for LLM tool calling and log ingestion.
2. Ingest raw subcontractor daily reports across {{subcontractor_trades}} at {{site_name}}.
3. Query the schedule baseline in {{schedule_baseline_tool}} to extract current milestones, dependencies, and float values.
4. Execute an LLM reasoning node to parse unstructured trade notes, extracting reported crew sizes, equipment downtime, and work completed.
5. Compare extracted work velocity against baseline milestones to compute schedule variance in hours.
6. Evaluate if variance exceeds {{delay_threshold_hours}} on active critical path activities.
7. Construct a standardized JSON payload detailing root cause, trade impact, and suggested mitigation steps for {{general_contractor}}.
8. Dispatch an asynchronous HTTP POST request to {{notification_webhook}} and log audit telemetry.

Constraints

• MUST produce fully runnable, modular Python code using modern type hints and error handling.
• MUST NOT hardcode API credentials; consume environment variables instead.
• Logic MUST explicitly separate critical path impacts from non-critical float consumption.
• Include inline documentation explaining fallback error states for corrupted daily logs.

Output format

• Code Header and Imports block
• Data Schema and Typed Models (Pydantic)
• Core Agent State Machine / Tool Execution functions
• Orchestration Runner and Webhook Dispatcher
• Deployment and Environment Variable Execution Guide

Self-review

• Verify that all six variables ({{site_name}}, {{general_contractor}}, {{schedule_baseline_tool}}, {{delay_threshold_hours}}, {{notification_webhook}}, {{subcontractor_trades}}) are actively utilized.
• Ensure script logic contains clear exception handling for failed schedule tool queries.
• Check that output strictly adheres to runnable script requirements without pseudo-code placeholders.