{"id":"com.twmarketdata/tw-market-data","slug":"com-twmarketdata-tw-market-data","title":"Tw Market Data","description":"Taiwan stock market data (TWMD): official-source, point-in-time-safe datasets via read-only tools.","version":"1.28.1","publisher":{"kind":"domain","id":"twmarketdata.com","label":"twmarketdata.com","url":"https://twmarketdata.com"},"website":"https://twmarketdata.com","repository":"","remotes":[{"type":"streamable-http","url":"https://mcp.twmarketdata.com/mcp","headers":[{"description":"TWMD API key (get one at https://twmarketdata.com). A key-free trial tier is available for sample tickers.","isRequired":true,"isSecret":true,"name":"X-API-Key"}]}],"packages":[],"published":"2026-07-22","updated":"2026-07-22","status":"online","auth":"","latency_ms":1463,"checked":"2026-11-19","uptime":100,"checks":3,"tools":[{"name":"list_datasets","title":"List datasets","description":"List available Taiwan-market datasets (discovery entry point). Returns id / 中文名 /\n    category / tier / one-line description for each. Use this first to find the right data.\n\n    Args:\n        category: optional, e.g. 'chip'(籌碼) 'fundamental'(基本面) 'price'(行情) 'macro'(總經)\n                  'relation'(關聯/產業鏈) 'derivatives'(期權) 'event'(事件) 'rag_text'(文本).\n        tier: optional minimum plan: 'free' '","read_only":true},{"name":"describe_dataset","title":"Describe a dataset's semantics","description":"FULL semantics of one dataset: grain (what a row is), field meanings+units, ★TIME-CORRECTNESS\n    rules (knowledge_time_field / point_in_time_safe — read before backtesting), relations for\n    cross-table reasoning, agent_hints (when to use), quant_use (which factors). Args: dataset_id.","read_only":true},{"name":"query_dataset","title":"Query a dataset (PIT-safe)","description":"Query rows with built-in look-ahead protection.\n\n    ★ POINT-IN-TIME: pass `as_of` (YYYY-MM-DD) for backtesting/agent-learning. For non-point-in-time-safe\n    datasets (fundamentals, monthly_revenue, dividend_policy…) rows are filtered by DISCLOSURE date <= as_of,\n    so the agent only sees what was public at that moment. Omit as_of only for present-day lookups (warned).\n\n    ★ IF A VALUE IS IN `c","read_only":true},{"name":"find_related","title":"Find related datasets and tickers","description":"Traverse the knowledge graph for cross-table / supply-chain reasoning.\n    - dataset_id: returns join-able datasets (+why) to plan multi-table analysis.\n    - ticker: returns its industry value-chain node + peers in the same node (supply-chain reasoning).\n    Args: dataset_id (e.g. 'equity_daily_prices') and/or ticker (e.g. '2330').\n    ","read_only":true},{"name":"search_filings","title":"Search filings and announcements","description":"Semantic search over MOPS filings and announcements — ⚠️ NOT YET AVAILABLE (語意檢索尚未提供/建置中).\n\n    ★ STATUS, READ FIRST: the semantic index has not been built. Every call currently returns NO\n    rows, with `corpus.index_is_empty: true` and `corpus.service_status: \"not_yet_available\"`.\n    Tell the user this capability is not available yet. Do NOT present it as a working feature,\n    and do NOT read ","read_only":true},{"name":"explain_provenance","title":"Explain where a value came from","description":"Where did this number come from, and when could anyone have known it? (WP-6)\n\n    ★ THIS IS THE GROUNDING TOOL. Every other tool's answer is supposed to be expandable through\n    this one: given a ticker, a field and a date, it names the dataset that serves that field, the\n    knowledge column, and WHICH as_of rule applies on that date. Use it whenever you are about to\n    state a number as fact.\n","read_only":true},{"name":"get_inclusion_proof","title":"Get an inclusion proof","description":"Prove a row was in the snapshot TWMD published — and check it yourself.\n\n    Returns the Merkle sibling path, the signed root, and the checkpoint it belongs to. It returns\n    the PATH rather than a yes/no on purpose: a service that answers \"yes, it is included, trust me\"\n    is the opposite of verifiable. Recompute the root from the leaf and the path; the verifier is\n    ~30 lines and is written ","read_only":true},{"name":"cite_this","title":"Cite a served number","description":"Produce a bibliographic citation for TWMD data — APA, BibTeX, and a re-verifiable token.\n\n    ★ FOR PAPERS, REGULATORY FILINGS AND ANYTHING A REVIEWER WILL RE-CHECK LATER. A dataset is\n    corrected, backfilled and re-run. Three years from now a reviewer opening our API sees different\n    numbers than the\n    paper, and nobody — author, reviewer, or us — can tell whether the data changed or the au","read_only":true},{"name":"read_primary_text","title":"Read the primary source text","description":"Read the FULL TEXT of filings and announcements — with proof links and a knowledge cutoff.\n\n    ★ NOT `search_filings`. That one ranks passages by similarity and hands you fragments; this hands\n    you whole documents so you can read what was actually said and where it sat in the filing.\n    Similarity is not importance, and a fragment cannot show you its own context.\n\n    ★ POINT-IN-TIME: pass `a","read_only":true},{"name":"ask","title":"Ask in natural language","description":"Answer a plain-language question in Taiwanese-market vocabulary, sentence by sourced sentence.\n\n    ★ FOR BEGINNERS WHO DO NOT KNOW WHICH DATASET THEY WANT. Ask \"PBR 是什麼\" or \"什麼叫漲跌停\"\n    in ordinary words; routing happens on our side. `describe_dataset` explains a table you already\n    named, `search_filings` digs through company disclosures, and `query_dataset` returns rows —\n    this one turns a","read_only":true},{"name":"chart","title":"Chart a series","description":"Turn rows you already fetched into a Vega-Lite drawing your chat window can render.\n\n    ★ IT DRAWS; IT DOES NOT FETCH. Hand it the output of `query_dataset` — this tool never touches\n    the database, so it cannot bypass the `as_of` filter those rows were selected under. A plotting\n    tool that fetched its own numbers would be a second data path, and a second path eventually\n    disagrees with t","read_only":true},{"name":"screen","title":"Screen the market","description":"Post-process rows already returned by query_dataset; without rows it returns a clear hint.\n\n    Turn a spoken shortlist description into explicit numeric cut-offs, and show the cut-offs.\n\n    ★ THE THRESHOLDS COME BACK WITH THE SHORTLIST. \"低本益比\" becomes `per < 15`, and that 15 is\n    printed in `applied[]` so you can disagree with it. A filter that hands over thirty names\n    without saying where ","read_only":true},{"name":"query_regime","title":"Query the macro regime","description":"The Taiwan business-cycle light (NDC monitoring indicator) as a monthly series.\n\n    ★ THIS IS A REVISED FIGURE, NOT A POINT-IN-TIME ONE. Every response carries\n    `revision_basis: as_revised`. Our source holds exactly one row per month — the CURRENT value,\n    not the value as first published — and it records no publication date.\n\n    ★ `as_of` IS REFUSED, AND THE REFUSAL IS THE POINT. There is ","read_only":true},{"name":"calendar","title":"Upcoming and past events","description":"Sort corporate dates into what is still ahead and what has already passed.\n\n    ★ TWO DATES, NOT ONE. What is \"upcoming\" is decided by the date the event HAPPENS; `as_of`\n    filters on the date it was ANNOUNCED. An ex-dividend declared on 2026-08-01 for 2026-09-15 is\n    both already known and still ahead on 2026-08-10. Collapsing the two fields either hides every\n    future date or reports last ","read_only":true},{"name":"compare","title":"Compare tickers","description":"Lay two to five named companies side by side on the same measures, gaps marked as gaps.\n\n    ★ AN ABSENT FIGURE STAYS ABSENT, AND THE COMPANY STAYS ON THE TABLE. A blank cell is reported\n    as `available: false`, never filled with a zero, a previous period, or by quietly dropping the\n    column. Dropping is the worst of the three: it converts \"we do not hold this measure for that\n    company\" int","read_only":true},{"name":"get_code_example","title":"Get a code example","description":"Emit a copy-pasteable HTTP snippet wired to the real endpoint, header and parameter names.\n\n    ★ FOR WRITING YOUR OWN CLIENT, NOT FOR GETTING DATA. Every other tool here answers a question;\n    this one hands you source code so your program can ask it directly over HTTPS. Nothing is\n    fetched and no rows come back.\n\n    ★ THE CONSTANTS ARE READ OUT OF THE SERVER, NOT REMEMBERED. Base URL, the `","read_only":true},{"name":"run_recipe","title":"Run a research recipe","description":"Replay a saved multi-step routine over rows you fetched, with every step listed.\n\n    ★ THREE ROUTINES: `momentum_scan` (rank by a return column), `earnings_surprise`\n    (actual versus estimate), `dividend_capture` (which ex-dates are still ahead).\n\n    ★ A SAVED ROUTINE IS NOT A TRADING VIEW. The names are conventional labels for well-known\n    sequences; what runs is arithmetic over rows you su","read_only":true},{"name":"try_sample","title":"Try a free sample (no key)","description":"Hand an unregistered caller a short taste of an open dataset, plus where to unlock the rest.\n\n    ★ WHAT AN ACCOUNTLESS CALLER GETS INSTEAD OF A BARE REFUSAL. Somebody arriving through a chat\n    connector with no plan would otherwise meet a flat rejection, which the host model relays as\n    \"this service turned you down\".\n\n    ★ WHAT IS ACTUALLY FREE, STATED CONCRETELY. The reference resources re","read_only":true},{"name":"run_backtest","title":"Run a backtest","description":"Submit a point-in-time backtest. NON-BLOCKING: returns a run_id at once; poll get_backtest.\n\n        `job_status` tells you where it is: `cached` (this exact backtest was already computed — the\n        stored result is returned immediately), `done` (it finished within `wait_seconds`), or\n        `queued` / `running` (call `get_backtest(run_id)` until it is `done` or `failed`). The same\n        spe","read_only":false},{"name":"get_backtest","title":"Get a backtest result","description":"Retrieve a previous backtest by run_id — the full record, including why it was rejected.\n\n        Only runs in YOUR namespace are visible; a run_id belonging to someone else is simply not\n        found. Args: run_id (the `twmd_bt_…` handle returned by run_backtest).\n        ","read_only":true},{"name":"replay_backtest","title":"Replay a backtest","description":"Re-run a stored backtest and report whether it still produces the same numbers.\n\n        Same spec, same `as_of`, same data questions. If the numbers moved, either the engine\n        version changed or the underlying data was restated — both are reported, neither is\n        smoothed over. Args: run_id.\n        ","read_only":true},{"name":"list_backtests","title":"List backtests","description":"Browse an INDEX of your past backtest runs — ids and headline metrics only, no re-execution.\n\n        Use when you want to find a run whose id you have forgotten. It never re-computes anything:\n        `run_backtest` executes a new one, `get_backtest` opens a single record in full, and\n        `replay_backtest` re-derives one to check reproducibility. This is the catalogue, not any of\n        thos","read_only":true},{"name":"memory_save","title":"Save to agent memory","description":"Remember something, with its sources and its knowledge time.\n\n        Nothing is ever overwritten: saving a `factor_def` or `watchlist` under an existing key\n        SUPERSEDES the previous version (both rows survive, so \"what did I believe in June?\" stays\n        answerable), and saving identical content twice is a no-op rather than a duplicate.\n\n        Args:\n            kind: 'query' | 'factor_","read_only":false},{"name":"memory_search","title":"Search agent memory","description":"Recall your own memories — hybrid (semantic + exact-term), with provenance attached.\n\n        Every result carries where it came from (`source_query_ids`, replayable), when it was\n        believed (`valid_from`/`valid_to`) and what knowledge time it is about (`as_of`), plus a\n        `recall` block stating which model and which filters produced the answer.\n\n        Args:\n            query: what yo","read_only":true},{"name":"memory_get_watchlist","title":"Get the watchlist","description":"Read the tickers on one named watchlist, as it stands right now.\n\n        A curated roster you maintain — distinct from `memory_search`, which digs through everything\n        you ever recorded, and from `list_alerts`, which is about price triggers rather than symbols\n        you are following. Args: key — the roster's name, default 'default'.\n        ","read_only":true},{"name":"memory_replay_query","title":"Replay a cited query","description":"Re-run a remembered query by its `twmd_q_…` id, through the read API's own replay store.\n\n        This is what makes a recalled finding checkable: the memory says where it came from, and\n        this fetches that same data again. It never re-executes the query by another route — two\n        implementations of \"replay\" would be two answers to a question whose whole value is having\n        one.\n\n   ","read_only":true},{"name":"run_research","title":"Run a research report","description":"Run a multi-agent research pass and return a structured, sourced report.\n\n        Six roles run in order — data analyst, factor researcher, backtest engineer, risk officer,\n        portfolio manager, compliance officer. Each step's output carries the `query_ids` behind it;\n        risks are reported alongside results, not beneath them; and anything the run could not do is\n        listed as a limit","read_only":false},{"name":"get_research","title":"Get a research report","description":"Retrieve one of YOUR previous research reports. Others' runs are simply not found.\n\n        Args: research_id — the identifier `run_research` returned when it started that run.\n        ","read_only":true},{"name":"list_pending_actions","title":"List pending actions","description":"Financial actions proposed by your research runs that are waiting for a human decision.\n\n        Nothing here has been executed or ever will be by this system. These are proposals.\n        ","read_only":true},{"name":"approve_action","title":"Approve a pending action","description":"Record a HUMAN's approval of a proposed action.\n\n        This writes an audit record naming who approved what, and when. It does NOT execute the\n        action: TWMD has no order or funds path, by design. Execution, if any, happens elsewhere and\n        is performed by a person.\n\n        Args:\n            action_id: from `list_pending_actions` or a report's `proposed_actions`.\n            approver","read_only":false},{"name":"list_alerts","title":"List price alerts","description":"Show the price-trigger rules you have armed, and whether each is still armed.\n\n        A read-only inventory of thresholds you asked to be watched — it arms nothing and cancels\n        nothing (`set_price_alert` arms, `delete_alert` cancels). Another customer's triggers are\n        simply not visible here.\n        ","read_only":true},{"name":"set_price_alert","title":"Set a price alert","description":"Leave a standing instruction: tell me when this symbol crosses this price.\n\n        The alert OUTLIVES this conversation. It is evaluated against official daily closes by a\n        resident agent and delivered to your realtime stream and to any webhook endpoints you have\n        registered — signed, retried, and de-duplicated so one crossing is one notification.\n\n        This is a NOTIFICATION, no","read_only":false},{"name":"delete_alert","title":"Delete a price alert","description":"Cancel one armed price trigger permanently, by its rule id.\n\n        Disarms a single watch so it will not fire again — the opposite of `set_price_alert`,\n        and unlike `list_alerts` it changes state rather than reporting it. Cancellation is\n        irreversible: re-arming means creating a fresh trigger. Naming somebody else's rule id\n        cancels nothing at all. Args: rule_id.\n        ","read_only":false},{"name":"risk_assess","title":"Assess portfolio risk","description":"Measure a portfolio you state against limits you state, on official point-in-time prices.\n\n        Reports concentration and peak-to-trough drawdown, and NAMES every position it could not\n        price rather than quietly assessing the rest — an assessment covering 60% of a portfolio\n        without saying so is worse than none.\n\n        Any breach produces a PROPOSAL (e.g. \"reduce 2330\") that req","read_only":true},{"name":"agent_activity","title":"Agent activity log","description":"What YOUR agents have actually done, from the durable audit trail.\n\n        Every resident agent records what it did and ON WHAT BASIS — the rule and the two closes\n        behind an alert, the coverage and limits behind a risk finding, the run_id and declared trial\n        count behind a factor verdict. This is that trail, and it survives deploys.\n\n        Reports which agents have recorded NOTHI","read_only":true},{"name":"list_factor_findings","title":"List factor findings","description":"Verdicts from the overnight factor search on YOUR namespace — including the rejections.\n\n        The rejections are returned deliberately. A research log that keeps only the winners is the\n        highlight reel overfitting lives in, and the acceptance RATE is the number that tells you\n        whether the anti-overfitting gate is doing its job: a search that accepts most of what it\n        tries h","read_only":true},{"name":"company_health_check","title":"Company health check","description":"一檔股票的誠實體檢:成長、獲利品質、估值、籌碼、紅旗,每個數字可驗證。\n\n        ★ 這是**事實整理**,不是投資建議、不是預測、不是目標價。輸出帶\n        `does_not_prove`,轉述時請一併保留。\n\n        ★ POINT-IN-TIME:`as_of` 逐資料集尊重揭露時差 —— 月營收用法定截止日推得的\n        估計時點,財報用出表日,估值是當日。不給 `as_of` 就用「現在可知的最新」,\n        **不是今天**。\n\n        ★ 沒看到的東西會列在 `cannot_see`,並說明是「查了沒有」「查詢失敗」還是\n        「散佈權未取得」—— 三者不同。⚠️ 不要替 `cannot_see` 裡的項目估算或補值。\n\n        ★ `checked_and_clear` 是「查過而且確認沒有」(例如近一年無","read_only":true},{"name":"positioning_read","title":"Read institutional positioning","description":"籌碼結構:法人分項、融資融券、借券使用率、大戶集中度、董監質押。\n\n        ★ 事實整理,不是投資建議、不是預測、不是目標價。輸出帶 `does_not_prove`。\n        ★ POINT-IN-TIME:`as_of` 逐資料集尊重揭露時差;不給就用「現在可知的最新」。\n        ★ 看不到的在 `cannot_see`,並分「查了沒有 / 查詢失敗 / 散佈權未取得」。\n          ⚠️ 逐券商分點進出**永遠**在 cannot_see —— 散佈權未取得,不要替它估算。\n        ★ PRECONDITIONS:自己取數,rows 不需要先用 `query_dataset` 備好;requires 一個\n          已識別的 session。方案沒涵蓋的資料集以 not_licensed 進 cannot_see,不整支失敗。\n   ","read_only":true},{"name":"macro_regime_read","title":"Read the macro regime","description":"總經 Regime:景氣信號、利率與殖利率曲線、匯率、資金流。\n\n        ★ 這支講的是**整個市場**,不是任何一檔個股 —— 所以它不收 ticker。\n        ★ 事實整理,不是投資建議、不是預測。輸出帶 `does_not_prove`。\n        ★ ⚠️ `macro_worldbank` 類的來源會**回溯修訂**歷史年度:今天讀到的舊年度數字\n          和當時看到的不是同一個,`pit_notes` 會標出來。\n        ★ Args:只有 `as_of`(選填)—— 它**不收 ticker**,因為主體是整個市場。\n        ★ Returns:同一個誠實信封;`subject` 固定是 \"TW_MARKET\"。\n        ★ 例如問「現在景氣循環在哪個位置」就是這一支,而不是問某一檔。\n        ★ PRECON","read_only":true},{"name":"supply_chain_read","title":"Read the supply chain","description":"供應鏈與同業:產業鏈位置、同業分組、轉投資、外銷訂單 vs 營收。\n\n        ★ 事實整理,不是投資建議、不是預測。輸出帶 `does_not_prove`。\n        ★ ⚠️ `company_peer_groups` 無日期欄:指定過去的 `as_of` 時,回的是**現況**分組,\n          `pit_notes` 會標「非當時分類」。\n        ★ ⚠️ 質化的客戶/供應商關係與分部營收在年報附註,那條來源**尚未建置** ——\n          固定列在 `cannot_see`(status=no_loader),不是這次查失敗。\n        ★ Args:`ticker`(例如 \"2330\")、`as_of`(選填)。\n        ★ Returns:誠實信封;外銷訂單與營收的背離會 listed in `concerns`。\n   ","read_only":true},{"name":"risk_read","title":"Read risk events","description":"事件與市場結構風險:隱含波動率、注意處置、裁罰訴訟、放空限制、因子。\n\n        ★ ⚠️ **部位集中度與回撤不在這裡** —— 那是 `risk_assess` 的職責。重算一份會產生\n          第二個答案,而兩個都帶著我們的名字。\n        ★ `checked_and_clear` 是「查過而且確認沒有」(例如近一年無裁罰),和\n          `cannot_see` 的「不知道」是兩回事。\n        ★ 事實整理,不是投資建議、不是預測。輸出帶 `does_not_prove`。\n        ★ Args:`ticker`(例如 \"2330\")、`as_of`(選填)。\n        ★ Returns:誠實信封;事件風險 listed in `concerns`,查過確認沒有的\n          listed in `checked_a","read_only":true}],"server_info":{"name":"tw-market-data","version":"1.28.1"},"category":"data","tags":["memory","finance","research"],"kind":"remote","source":"registry","submitted":{},"first_party":true,"domain_rank":null,"score":100,"indexable":true,"first_seen":"2026-07-22","stars":0,"badges":[],"sponsor":null,"page":"https://agentsindex.org/agents/com-twmarketdata-tw-market-data/"}