$npx -y skills add Power-Agent/PowerSkills --skill egretProgressive-disclosure workflow for Egret market and operations optimization. Use whenever the user wants to run DC-OPF, AC-OPF, or unit commitment through PowerMCP — even when they just say "what's the cheapest dispatch", "commit the units", "check for congestion", or "is this s
| 1 | # Egret workflow |
| 2 | |
| 3 | Start with the cheapest model that can answer the question. Do not jump into unit commitment when a single-snapshot OPF will do. |
| 4 | |
| 5 | ## Default tool ladder |
| 6 | 1. `solve_dc_opf(case_file, solver, return_results)` for fast feasibility and congestion screening. |
| 7 | 2. `solve_ac_opf(case_file, solver, return_results)` when voltage or reactive realism matters. |
| 8 | 3. `solve_unit_commitment_problem(case_file, solver, mipgap, timelimit)` only after the horizon, reserve logic, and commitment assumptions are clear. |
| 9 | |
| 10 | ## Working rules |
| 11 | - Confirm the case file format, solver, and time horizon before solving. |
| 12 | - Use DC OPF first unless the question explicitly depends on AC feasibility. |
| 13 | - Treat UC as a higher-cost study that needs a clear operational objective. |
| 14 | - Validate promising dispatch fixes in an AC tool when voltage limits matter. |
| 15 | |
| 16 | ## Escalation triggers |
| 17 | Quote the binding constraint or infeasibility cause rather than saying "it failed". |
| 18 | |
| 19 | | Observation | Escalate to | |
| 20 | |---|---| |
| 21 | | OPF or UC infeasible, reserve-short, or implausibly expensive | `operations-planning-mitigation` | |
| 22 | | AC-OPF bus voltage outside [0.95, 1.05] pu | `voltage-violation-mitigation` | |
| 23 | | A branch limit binds persistently across the horizon | `thermal-overload-mitigation` | |
| 24 | |
| 25 | ## Deliver |
| 26 | - The case, solver, and study type. |
| 27 | - The limiting constraints or binding tradeoffs. |
| 28 | - Whether the next step is AC validation, UC escalation, or mitigation. |