Agenten erreichen Infrastruktur nur über diese Tools. Kein Tool schreibt einen Sollwert. Jedes Tool trägt die neun Anforderungen unten; „specified“ heißt: Vertrag fixiert, Implementierung existiert noch nicht.
schemaauthenticationauthorizationdeterministic validationtimeoutprovenanceloggingfailure behaviourtest coverage
inspect_grid_state()Snapshot of the twin at a timestamp: buses, branches, assets, measurements, active alarms.
tenant:read2000 ms{ tenant, network_id, at?: iso8601, depth?: 1|2|3 }{ snapshot_id, state, provenance }retrieve_asset()Asset record, parameters, limits, connection point, health features.
tenant:read1000 ms{ tenant, asset_id }{ asset, limits, provenance }run_power_flow()AC (Newton) or DC power flow on a snapshot with optional modifications.
tenant:simulate10000 ms{ snapshot_id, method: ac|dc|pinn, modifications?: Action[] }{ result_id, voltages, flows, converged, iterations, solver, latency_ms }run_contingency()N-1 / N-k screening over a snapshot.
tenant:simulate30000 ms{ snapshot_id, set: n-1|list, limit_margin_pct }{ violations[], worst_case, latency_ms }run_forecast()Probabilistic forecast for load, PV, price or constraint at a node.
tenant:read5000 ms{ tenant, target, horizon_h, quantiles[] }{ forecast_id, series[quantile], model_version, backtest_error }simulate_action()Roll the twin forward under a candidate action set.
tenant:simulate20000 ms{ snapshot_id, actions: Action[], horizon_min }{ sim_id, trajectory, constraint_report }calculate_flexibility()Available up/down flexibility per asset and aggregate over a window.
tenant:read5000 ms{ tenant, asset_ids[], window }{ per_asset[], aggregate, assumptions[] }optimize_dispatch()MILP dispatch for BESS / DER portfolio against price, constraints and SoC.
tenant:simulate60000 ms{ tenant, portfolio, objective, horizon_h, constraints }{ plan_id, schedule, objective_value, baseline_value, gap, solver_status }validate_constraints()Deterministic pass/fail of a plan against physics and policy, with margins.
tenant:simulate10000 ms{ plan_id | sim_id }{ pass, violations[], margins }request_authorization()Submit a validated plan for human or envelope authorization. The only tool that leaves the sandbox — and it produces a request, not a setpoint.
tenant:request3000 ms{ plan_id, sim_id, validation_id, explanation_id, requested_window }{ request_id, status: pending|approved|denied, expires_at }generate_explanation()Plain-language and structured explanation of a plan: assumptions, trade-offs, what was rejected.
tenant:read15000 ms{ plan_id, audience: operator|manager|auditor, lang: en|de }{ explanation_id, text, structured }create_incident()Open an incident with evidence attached (snapshots, sims, forecasts).
tenant:write:incident3000 ms{ tenant, severity, title, evidence_ids[] }{ incident_id }generate_audit_record()Append a hash-chained record of who asked what, what was computed, what was decided.
system1000 ms{ actor, tool, inputs_hash, outputs_hash, decision? }{ record_id, chain_hash }