openclaw-ubem

Generate per-building PDF calibration reports or the campus-wide summary PDF for the Applied Energy paper. Use when asked to generate a report, write up findings, create a calibration log, or produce paper-ready results.

Generate or refresh the NUS building registry (building_registry.json) by scanning all IDF files. Use when IDF files change, when a new building is added, or when asked about building metadata, floor areas, zone counts, or IDF parameters. Also used for pre-flight registry validation before simulation. Owned by SimulationAgent (Forge).

Run EnergyPlus simulations for NUS campus buildings and parse monthly energy output. Use when SimulationAgent is asked to simulate one or all buildings, check calibration status, or re-run after IDF changes.

Run targeted EnergyPlus intervention simulations for NUS buildings. Patches IDF parameters (cooling setpoint, lighting, equipment loads), re-simulates, and reports actual kWh/carbon savings vs baseline. Use when asked to propose or simulate interventions for a folder or set of buildings.

Format and route NUS energy system notifications to Slack. Two message types only — calibration approval requests and final intervention results. All other pipeline events stay in agent logs.

Run and manage the NUS Energy Assistant Slack bot server (Socket Mode). Handles inbound @Energy_assistant mentions and DMs in Slack, routes questions to QueryAgent, and posts answers in thread. Use when starting/stopping the server, checking its status, or diagnosing Slack query failures.

Fetch live and historical weather data from NUS onsite stations and the data.gov.sg API, validate sensor readings, and build site-calibrated EPW files for EnergyPlus simulations. Use when asked for current conditions, observed weather data for a past month, a calibrated EPW for Forge, or weather context for anomaly diagnosis.

Carbon scenario engine for NUS buildings. Reads a building's parsed monthly CSV and IDF file, extracts an energy fingerprint, scores 7 interventions, assembles ranked reduction scenarios, and estimates Singapore-market retrofit costs (SGD) for shallow / medium / deep tiers. Use when asked for carbon scenarios, reduction potential, intervention analysis, cost estimates, payback periods, or net-zero pathways for any NUS building.

Core identity and values for the NUS Energy Assistant. Always active. Defines who I am, what I know about NUS buildings, calibration rules, carbon targets, and how I communicate with the facilities team and orchestrator.

Answer ad-hoc questions about NUS campus energy performance by reading pipeline output data. Use for questions about MAPE, calibration status, energy intensity, carbon footprint, BCA benchmarks, cost savings, and building comparisons. Always ground answers in real data files.

Propose and apply IDF parameter changes to improve calibration for NUS buildings. Use when ASHRAE metrics fail, when asked to fix setpoints or infiltration, or when the model over/under-predicts. ALWAYS requires human approval before writing any IDF change.

Core identity and values for the NUS Energy Assistant. Always active. Defines who I am, what I know about NUS buildings, calibration rules, carbon targets, and how I communicate with the facilities team and orchestrator.

Diagnose root causes for NUS buildings that fail ASHRAE calibration thresholds. Use when a building's CVRMSE > 15% or NMBE > ±5%, when asked why a building is over/under-predicting, or when proposing parameters for RecalibrationAgent.

Core identity and values for the NUS Energy Assistant. Always active. Defines who I am, what I know about NUS buildings, calibration rules, carbon targets, and how I communicate with the facilities team and orchestrator.

Compare simulated energy against canonical NUS ground-truth meter data and compute ASHRAE Guideline 14 calibration metrics (CVRMSE, NMBE) for all GT buildings across 5 archetypes. Produces per-building verification CSVs and per-archetype calibration aggregates used to propagate parameters to non-GT buildings. Use when asked about calibration status, ASHRAE pass/fail, or how well the model matches reality.

Parse EnergyPlus .mtr output into monthly energy CSV for a NUS building. Use after simulation completes, or when asked to extract monthly kWh, cooling energy, or carbon from simulation results.

Answer natural language questions about NUS campus energy performance, carbon footprint, BCA benchmarks, building comparisons, and cost savings. Use for any question about what the data shows — EUI rankings, carbon totals, savings estimates, calibration summaries.

Core identity and values for the NUS Energy Assistant. Always active. Defines who I am, what I know about NUS buildings, calibration rules, carbon targets, and how I communicate with the facilities team and orchestrator.

Daily pipeline orchestration for the NUS campus energy management system. Coordinates all 8 specialist agents in the correct 7-phase sequence. Use when running the full daily pipeline, triggering a partial re-run after recalibration, or deciding which buildings need action.