> ## Documentation Index > Fetch the complete documentation index at: https://octo-98895abd.mintlify.site/llms.txt > Use this file to discover all available pages before exploring further. # Sheets and Excel > Pull SILO data into Google Sheets and Excel as CSV. What the server does was measured live; the spreadsheet snippets are marked where they were not run. ### Overview The API speaks CSV. Ask for it with `Accept: text/csv` and every view and every function answers with a header row and comma-separated values, straight into a spreadsheet. The catch is headers. A spreadsheet formula like Google Sheets' `IMPORTDATA` cannot set any, and without `Accept: text/csv` the server answers JSON. So the working recipes are the ones that can set a header: **Apps Script** in Sheets, **Power Query** in Excel. Everything under "What the server does" was observed on 2026-09-24 against the live API with the curl commands shown. The Apps Script and Power Query snippets were **not executed**: they are written against those observations, and are marked as such. ### What the server does Base URL and key are the ones on [Quickstart](/api-docs/quickstart): ```bash theme={null} export SILO_URL="https://zcjbtpxuhdekpwcxmepn.supabase.co/rest/v1" export SILO_ANON_KEY="sb_publishable__yfFQsykAglrvc9GS6_PYw_B24ex437" ``` | Request | Observed | | - | - | | View, `Accept: text/csv` | `200`, `text/csv; charset=utf-8`, header row then data | | Function via `POST /rpc/…`, `Accept: text/csv` | `200`, `text/csv` | | Function via `GET /rpc/…?p_…=…`, `Accept: text/csv` | `200`, `text/csv` | | Key as `?apikey=` in the URL, no `apikey` header | `200`: the gateway accepts it | | No `Accept-Profile` header | `200`: `api` is the default schema | | No `Accept` header, or `Accept: */*` | `200`, **`application/json`** | | No key anywhere | `401`, `No API key found in request` | | A function over its row cap, CSV requested | `400`, **JSON** error body (`22023`) | | A view with `limit=5000`, CSV requested | 1,000 rows; `Content-Range: 0-999/*` | | `equities?ticker=eq.PETR4&limit=5000` plus `Prefer: count=exact`, CSV | `206`, `Content-Range: 0-999/1925` | A view, as CSV: ```bash theme={null} curl -s "$SILO_URL/equities?ticker=eq.PETR4&select=ticker,trade_date,close,volume&order=trade_date.desc&limit=5" \ -H "apikey: $SILO_ANON_KEY" \ -H "Accept: text/csv" ``` ``` ticker,trade_date,close,volume PETR4,2026-09-23,49.600000,2157789085.00 PETR4,2026-09-22,48.350000,2350910695.00 PETR4,2026-09-21,48.000000,1383164736.00 PETR4,2026-09-18,48.500000,2554613048.00 PETR4,2026-09-17,48.610000,1574576494.00 ``` A function, as CSV. `POST` with a JSON body, or `GET` with the arguments in the query string; both return the same CSV: ```bash theme={null} curl -s -X POST "$SILO_URL/rpc/quote_latest" \ -H "apikey: $SILO_ANON_KEY" \ -H "Content-Type: application/json" \ -H "Accept: text/csv" \ -d '{"p_ticker":"PETR4"}' curl -s "$SILO_URL/rpc/quote_latest?p_ticker=PETR4&apikey=$SILO_ANON_KEY" \ -H "Accept: text/csv" ``` `quote_history` returns one JSON object per session holding only the fields you select (default `ticker`, `trade_date`, `close_adj`), so as CSV it is a single column of JSON. For a price series in a sheet, read the typed views (`equities`, `units`, raw closes) as below, or call `quote_history` as JSON. The key in the URL, with no headers at all, returns JSON: ```bash theme={null} curl -s "$SILO_URL/equities?ticker=eq.PETR4&select=ticker,trade_date,close&limit=2&apikey=$SILO_ANON_KEY" # [{"ticker":"PETR4","trade_date":"2019-01-02","close":24.060000}, ...] ``` **The 1,000-row cut is invisible in a spreadsheet.** A view asked for 5,000 rows returned 1,000, and the only sign was the `Content-Range` response header, which no spreadsheet shows. The sheet simply ends. Filter to fewer than 1,000 rows, or check the count as the Apps Script below does. Functions refuse instead of cutting (the `400` above), which is safer, but the refusal arrives as JSON. See [Paging without losing rows](/api-docs/guides/pagination). ### Google Sheets: `IMPORTDATA` does not work `IMPORTDATA(url)` fetches a URL and splits it as CSV. It cannot set headers, so: * the key can go in the URL (`?apikey=…` worked, above); * the schema needs no header (`api` is the default); * but the format **cannot be chosen**, and without `Accept: text/csv` the server returned JSON in every request we made. What Accept header Google's fetcher actually sends was not observed. Unless it sends `text/csv`, `IMPORTDATA` gets JSON and splits it on commas into cells like `[{"ticker":"PETR4"`. We found no way to request CSV from the URL alone, so `IMPORTDATA` is **not a supported recipe**. ### Google Sheets: Apps Script custom function **Not executed.** Written against the server behaviour above, not run inside Google Sheets. `UrlFetchApp.fetch`, `Utilities.parseCsv` and custom functions are standard Apps Script; the details (header casing, number handling in your locale) are exactly where an untested snippet can be wrong. Extensions → Apps Script, paste, save: ```javascript theme={null} const SILO_URL = "https://zcjbtpxuhdekpwcxmepn.supabase.co/rest/v1/"; const SILO_KEY = "sb_publishable__yfFQsykAglrvc9GS6_PYw_B24ex437"; /** * A view as a table. =SILO("equities?ticker=eq.PETR4&select=ticker,trade_date,close&order=trade_date.desc&limit=30") * @customfunction */ function SILO(path) { const res = UrlFetchApp.fetch(SILO_URL + path, { headers: { apikey: SILO_KEY, Accept: "text/csv", Prefer: "count=exact" }, muteHttpExceptions: true, }); return siloTable_(res); } /** * A function as a table. =SILO_RPC("quote_latest", "{""p_ticker"":""PETR4""}") * @customfunction */ function SILO_RPC(name, jsonArgs) { const res = UrlFetchApp.fetch(SILO_URL + "rpc/" + name, { method: "post", contentType: "application/json", payload: jsonArgs || "{}", headers: { apikey: SILO_KEY, Accept: "text/csv" }, muteHttpExceptions: true, }); return siloTable_(res); } function siloTable_(res) { const code = res.getResponseCode(); if (code !== 200 && code !== 206) { throw new Error("SILO " + code + ": " + res.getContentText()); } // A view cut by the server's 1,000-row cap says so only here: "0-999/5321". // A smaller page you asked for with limit= is not an error. const headers = res.getHeaders(); const range = headers["Content-Range"] || headers["content-range"]; if (range) { const m = String(range).match(/^(\d+)-(\d+)\/(\d+)$/); const got = m ? Number(m[2]) - Number(m[1]) + 1 : 0; if (m && got === 1000 && Number(m[2]) + 1 < Number(m[3])) { throw new Error( "SILO cut the result at 1,000 of " + m[3] + " rows. Narrow the filter or page with offset.", ); } } const rows = Utilities.parseCsv(res.getContentText()); // Values arrive as text; turn plain numbers into numbers. return rows.map((r, i) => i === 0 ? r : r.map((v) => (/^-?\d+(\.\d+)?$/.test(v) ? Number(v) : v)), ); } ``` The key sits in the script, not in the sheet's cells. It is the shared publishable key printed on [Quickstart](/api-docs/quickstart), so this hides nothing, but anyone with edit access to the sheet can read it. ### Excel: Power Query **Not executed.** Written against the server behaviour above, not run in Excel. `Web.Contents` with `Headers`, `Csv.Document` and `Table.PromoteHeaders` are standard Power Query M. Data → Get Data → From Other Sources → Blank Query → Advanced Editor: ``` let Source = Web.Contents( "https://zcjbtpxuhdekpwcxmepn.supabase.co/rest/v1/", [ RelativePath = "equities", Query = [ ticker = "eq.PETR4", select = "ticker,trade_date,close,volume", order = "trade_date.desc", limit = "250" ], Headers = [ apikey = "sb_publishable__yfFQsykAglrvc9GS6_PYw_B24ex437", Accept = "text/csv" ] ] ), Csv = Csv.Document(Source, [Delimiter = ",", Encoding = 65001, QuoteStyle = QuoteStyle.Csv]), Promoted = Table.PromoteHeaders(Csv, [PromoteAllScalars = true]), Typed = Table.TransformColumnTypes( Promoted, {{"trade_date", type date}, {"close", type number}, {"volume", type number}}, "en-US" ) in Typed ``` For a function, pass a body; `Web.Contents` sends a `POST` when `Content` is set: ``` Source = Web.Contents( "https://zcjbtpxuhdekpwcxmepn.supabase.co/rest/v1/", [ RelativePath = "rpc/quote_latest", Headers = [ apikey = "sb_publishable__yfFQsykAglrvc9GS6_PYw_B24ex437", Accept = "text/csv", #"Content-Type" = "application/json" ], Content = Text.ToBinary("{""p_ticker"":""PETR4""}") ] ) ``` If Excel asks how to authenticate, choose **Anonymous**: the key travels in the header. The `"en-US"` culture on the type step matters. The server writes decimals with a dot (`49.600000`), and a workbook in a Brazilian locale reads a dot as a thousands separator. We did not test this in a pt-BR Excel; typing the columns with an explicit culture is the safe default either way. ### Key in the URL: the trade-off It works (`?apikey=…`, observed above), and it is the only way to authenticate from a tool that cannot set headers. Two honest caveats: * The key then appears in the sheet, the query, the browser history and any proxy log. It is the **shared publishable key**, already printed on these docs, so nothing secret leaks. A **user token** from sign-in is different: never put one in a URL. * It does not solve the format. A tool that cannot set `Accept` gets JSON, so the URL-key route is useful only to tools that parse JSON. ### Related Auth tiers, the 1,000-row cap, nulls. Three ways a response gets cut, and only one tells you. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.