Update to The Expected Value

{
  "description": "Research how one exact trading-card printing or sealed product's observed price changed over a date range, with marketplace, finish, currency and missing-day coverage kept explicit.",
  "included_files": [
    {
      "relative_path": "agents/openai.yaml",
      "size_in_bytes": 236
    }
  ],
  "name": "price-history-research",
  "skill_md_contents": "---\nname: price-history-research\ndescription: Research how one exact trading-card printing or sealed product's observed price changed over a date range, with marketplace, finish, currency and missing-day coverage kept explicit.\n---\n\nUse `search_card_prices` or `search_sealed_prices` to establish identity, then\n`summarize_price_history` for a bounded statistical summary. This is historical\nresearch, not a numerical forecast or an instruction to trade.\n\n- Cards require the exact game, history set code, TCGplayer product ID and\n  `sub_type_name` such as Normal or Foil. Keep collector number, frame/treatment\n  and finish distinct. Sealed products require the exact game, base set, model\n  and TCGplayer product ID. Do not choose a similarly named printing or deck.\n- Honor the user's range, marketplace and price basis. If no range is given,\n  use the last 30 calendar days through today and state those explicit dates.\n  Choose a single supported marketplace/basis; label any assumption. One call\n  covers 1 to 366 inclusive days. Broader research needs explicit windows; do\n  not average window medians or percentage changes into a false overall result.\n- Cardmarket card history supports the `price` basis in EUR. Other supported\n  combinations use `low` or `market`; preserve the returned currency. Never\n  substitute a market quote for a missing low quote or convert currencies\n  without a separately sourced rate and user request.\n- Report the actual first and last usable observation dates and values,\n  absolute/percentage change, minimum, maximum, median, and observed/requested\n  days. Distinguish the requested date range from observed endpoints. A single\n  observation cannot establish change, and a zero starting price has no finite\n  percentage change.\n- Check `coverage.complete` before calling the summary complete. If false,\n  narrow the range or report the limit. Missing days are not filled; conflicting\n  same-day observations are excluded and counted. Sparse coverage is part of\n  the result, not a claim of price stability.\n\nUse `card_price_history` or `sealed_price_history` for the underlying dated rows\nwhen the user wants details or a chart; follow pagination and do not invent\nintermediate points. Cite the source returned by the tools and identify the exact\nproduct. Keep current price research separate from historical observations.\n"
}