tlly

Attach short dated notes to a series (e.g. 'Policy rate raised to 4.5%'); shown as markers on charts. Single-line plain text, max 280 chars; one note per date (re-adding replaces it); each counts as one data point.

Create a hosted dashboard; returns its public URL. Dashboard spec: {"title": str, "description"?: str, "range"?: "1m"|"3m"|"6m"|"ytd"|"1y"|"2y"|"5y"|"10y"|"max", "theme"?: "auto"|"light"|"dark" (default mode; auto follows the viewer's system and viewers can switch), "layout"?: "auto"|"overview"|"detail" (only the default for panels without "display": overview makes them tiles, detail makes them charts, auto makes them tiles above 8 panels), "visibility"?: "public"|"private" (default public: anyone with the link; private: only members of this account signed in to tlly.dev, so embeds and shared links stop working for others), "panels": [{"title"?: str, "series": "name" | ["a","b"], "type"?: "line"|"area"|"bar"|"stat"|"gauge"|"bargauge"|"pie"|"heatmap"|"table", "unit"?: str, "note"?: str, "stale_after_days"?: num, "thresholds"?: {"red_above"?, "yellow_above"?, "yellow_below"?, "red_below"?}, "range"?: (as above, overrides the dashboard's), "section"?: str (starts a new titled section at this panel), "views"?: bool (show the Level / Y/Y % / Index buttons; default: shown unless the unit is already a rate like %, y/y or pp), "display"?: "tile"|"chart" (tile: a compact number with a sparkline, click for the full chart; chart: a card with the full chart; mix them freely, e.g. tiles for the many supporting numbers and charts for the few that need their history), "headline"?: bool (show as a big number at the top of the dashboard, under a status strip; use for the 3-6 most important), "width"?: "half"|"full", "scale"?: "linear"|"log", "id"?: str (stable panel id, e.g. pn_abc123XYZ012; keep existing ids when editing so alerts stay attached), "right_axis"?: [series names] (line, area and bar charts: draw these series against a second y axis on the right, for series of different sizes or units; at least one stays on the left), "right_unit"?: str (unit of the right axis; "unit" is the left's), "left_label"?/"right_label"?: str (≤40 chars, a title written vertically along the left / right y axis, e.g. "Policy rate"; right_label needs right_axis), "view"?: "level"|"yoy"|"index" (initial view; viewers can switch), "source"?: str (shown as "Source: …"; defaults to the job's host), "labels"?: [display name per series], "status_series"?: name (a red/yellow/green text series that sets the panel colour), "compare"?: "prev"|"3m"|"1y" (what the change is measured against), "period"?: "D"|"M"|"Q" (show dates as 2026-06 or 2026Q2), "value_names"?: {"green": "Hold", …} (display names for text values)}], "badge"?: "worst" (default: the header badge shows the worst panel) | "off" | panel index (0-based; the badge follows that panel), "shade"?: [{"from": date, "to": date, "label"?: str}] (up to 20 shaded periods on every chart)}. "type": "table" shows one row per series (up to 20): label, latest value, date and a status dot; its "status_series" is a list with one entry per row ("" for none). "area" is a line chart filled underneath; "gauge" is a dial of the latest value with the threshold zones, and "bargauge" a horizontal bar per series (up to 20, coloured by thresholds), both scaled by "min"?/"max"?: num (default 0 to a bit above the largest value or threshold); "pie" is a donut of the series' latest values (shares of a whole, up to 20); "heatmap" is a calendar of the first series' daily values (good for activity per day). Status colour is computed from the latest value of the panel's first series; a series of red/yellow/green labels (e.g. a derived status) colours itself. Notes may use {latest}, {date}, {period} and {change}, filled in from the panel's latest value. Annotations on a panel's series appear as markers. Viewers can pick a time range (1M to Max, year to date, or a custom from-to period), zoom, switch views and download CSV; ?embed=1 hides the page chrome and &panel=N shows one panel. Text limits: title ≤120, description ≤500, panel title ≤120, unit ≤20, note ≤280, section and source ≤80 characters, single line, plain text.

Schedule a fetch job. It runs once immediately and the result is returned. Job spec: {"series": name, "cron": "m h dom mon dow" (UTC, at most every 30 min), "request": {"url", "method"?: "GET"|"POST", "headers"?: {}, "body"?: obj|str, "pages"?: {"param": "page", "start"?: 0, "max"?: 1-10} (json with items: bump a query parameter until an empty items array) | {"cursor": "paging.next.after", "into": "after", "max"?: 1-10} (copy that JSON path from each response into the JSON body; stop when it is missing. A cursor that remains after max pages is an error, so a sum is never a partial total. HubSpot CRM search uses this. A json aggregate with no request.pages fails when the body still has another page, paging.next.after, has_more true, links.next, pageInfo.hasNextPage or another next cursor, instead of saving a partial total. Set pages. A response with no next marker is complete.)}, "extract": {"format": "json", "items"?: "path.to.array", "date"?: "field", "value": "field", "value_type"?: "number"|"text", "aggregate"?: "count"|"sum"|"mean"|"min"|"max" (records per date, e.g. hourly prices to a daily mean; needs items; all but count need value), "period"?: "D"|"M"|"Q", "where"?: [{"field": "path", "equals"|"not_equals": value}] (up to 5, with aggregate). Omit "date" with aggregate to snapshot every matching record as one point on the fetch's half hour. String numbers (HubSpot properties.amount) are summed.} | {"format": "json_columns", "dates": "path.to.array", "values": "path.to.array", "value_type"?} (two parallel arrays, e.g. Yahoo chart: timestamp + close) | {"format": "csv", "date"?: "COLUMN", "value": "COLUMN", "delimiter"?, "value_type"?} (omit date to stamp every row with the fetch's half hour; one value per half hour, so the last row wins. That does not sum the rows the way a JSON aggregate that omits date does. A history needs a Date column. Make a copy https://docs.google.com/spreadsheets/d/1Ty5X3LkY2unxTNGbkq36qDzuuYxQMsuywaTNyzL_pac/copy. Public CSV https://docs.google.com/spreadsheets/d/1Ty5X3LkY2unxTNGbkq36qDzuuYxQMsuywaTNyzL_pac/export?format=csv&gid=1746689811. Downloads /templates/google-sheets.csv and /templates/google-sheets.xlsx.) | {"format": "jsonstat2"} (SSB/Eurostat; filter the query to a single series), "transform"?: [{"op": "pct_change"|"diff"|"rolling_mean"|"rolling_sum"|"rolling_min"|"rolling_max"|"shift", "n": int} | {"op": "scale", "factor": num}], "keep"?: "all"|"last"}. "value_type": "text" stores labels (no transforms; see Data policy). Dates like 2026-09-30, 2026M08, 2026Q2/2026K2, 2026 or unix time are normalised to ISO period-end dates; date-times (2026-10-05T14:00+02:00) count on the day as written. A single JSON value with no timestamp (omit "date", or a date path the body doesn't have) is dated with the fetch's half hour: that is valid for a snapshot such as a price. A date that is present but not a date is an error. Do not invent a timestamp field. URLs and string bodies may use {today}, {yesterday}, {tomorrow}, {year}, {month}, {day} (UTC, filled in on every run) for sources that put the date in the URL. Private APIs: the URL, header values and string body may use {{secret:NAME}} for a secret the workspace owner stored (list_secrets shows the names and the one host each may be sent to); the value is filled in at fetch time and never shown, e.g. "headers": {"Authorization": "Bearer {{secret:STRIPE_KEY}}"}. A stored secret may be sent on GET, on a POST whose body is a GraphQL query, or on POST https://api.hubapi.com/crm/v3/objects/{deals,contacts,companies}/search only (HubSpot's read-only CRM search, JSON body; https://developers.hubspot.com/docs/api-reference/legacy/crm/search-the-crm). No other POST may carry a secret, and the secret is still sent only to the host it was saved for. If a needed secret is missing, ask the user to add it at /account under Data sources.

Create or replace a derived series, computed from other series every time it is read (always fresh, stores no points, counts toward the series limit). definition: {"op":"add"|"subtract"|"multiply"|"divide","a":series,"b":series} (matched by date) | {"op":"pct_change"|"diff"|"moving_average"|"rolling_min"|"rolling_max"|"shift","a":series,"n"?:1-1000} (shift: the value n points earlier) | {"op":"scale","a":series,"factor":number} | {"op":"resample","a":series,"to":"M"|"Q","how"?:"last"|"mean"|"sum"|"ffill"} (to month/quarter ends; ffill also fills gaps up to today) | {"op":"mean"|"sum","inputs":[2-20 series]} (dates all inputs have) | {"op":"status","a":series,"red"?:[rule],"yellow"?:[rule]} → text labels red/yellow/green (red if any red rule holds, else yellow if any yellow rule, else green); rule: {"series"?:name (default a; read as of each date of a),"above"|"below":number,"for"?:1-24 (holds for the last k points)} or {"all":[2-10 rules]}; at most 10 rules | {"op":"composite","inputs":[status series],"min_inputs"?:n} → 0-100 score (red 2, yellow 1, green 0) | {"op":"count","inputs":[status series],"value":"red"|"yellow"|"green","share"?:true (percent)}. Inputs may themselves be derived (up to 10 levels; one read may touch at most 200 stored series). Example: a yield-curve spread {"op":"subtract","a":"govt_10y","b":"tbill_3m"}.

Delete annotations of a series on the given dates, or all of them if dates is omitted.

Delete a dashboard (series data is kept).

Delete a job (its series data is kept).

Delete all points of a series, or remove a derived series' definition.

Read one integration playbook by id: what to track, the minimum credential and scopes, the exact tlly job, and gotchas. google-sheets and json-api also return a template field (Make a copy, public CSV, downloads). Make a copy https://docs.google.com/spreadsheets/d/1Ty5X3LkY2unxTNGbkq36qDzuuYxQMsuywaTNyzL_pac/copy. Public CSV https://docs.google.com/spreadsheets/d/1Ty5X3LkY2unxTNGbkq36qDzuuYxQMsuywaTNyzL_pac/export?format=csv&gid=1746689811. Downloads /templates/google-sheets.csv and /templates/google-sheets.xlsx. Public: no API key required. Call this before inventing an API URL or scope.

Read the latest points of a series.

Health check: failing jobs, stale or empty panels, dashboards, storage usage. Call this to verify everything works.

Storage usage against quotas (data points = values + annotations, series, dashboards, jobs) and the largest series. Check before large writes.

Fetch a job spec once and save the result, without scheduling it or using a job slot ("cron" is optional and ignored). Use for one-off dashboards whose data doesn't need to update; points count toward the data quota. Job spec: {"series": name, "cron": "m h dom mon dow" (UTC, at most every 30 min), "request": {"url", "method"?: "GET"|"POST", "headers"?: {}, "body"?: obj|str, "pages"?: {"param": "page", "start"?: 0, "max"?: 1-10} (json with items: bump a query parameter until an empty items array) | {"cursor": "paging.next.after", "into": "after", "max"?: 1-10} (copy that JSON path from each response into the JSON body; stop when it is missing. A cursor that remains after max pages is an error, so a sum is never a partial total. HubSpot CRM search uses this. A json aggregate with no request.pages fails when the body still has another page, paging.next.after, has_more true, links.next, pageInfo.hasNextPage or another next cursor, instead of saving a partial total. Set pages. A response with no next marker is complete.)}, "extract": {"format": "json", "items"?: "path.to.array", "date"?: "field", "value": "field", "value_type"?: "number"|"text", "aggregate"?: "count"|"sum"|"mean"|"min"|"max" (records per date, e.g. hourly prices to a daily mean; needs items; all but count need value), "period"?: "D"|"M"|"Q", "where"?: [{"field": "path", "equals"|"not_equals": value}] (up to 5, with aggregate). Omit "date" with aggregate to snapshot every matching record as one point on the fetch's half hour. String numbers (HubSpot properties.amount) are summed.} | {"format": "json_columns", "dates": "path.to.array", "values": "path.to.array", "value_type"?} (two parallel arrays, e.g. Yahoo chart: timestamp + close) | {"format": "csv", "date"?: "COLUMN", "value": "COLUMN", "delimiter"?, "value_type"?} (omit date to stamp every row with the fetch's half hour; one value per half hour, so the last row wins. That does not sum the rows the way a JSON aggregate that omits date does. A history needs a Date column. Make a copy https://docs.google.com/spreadsheets/d/1Ty5X3LkY2unxTNGbkq36qDzuuYxQMsuywaTNyzL_pac/copy. Public CSV https://docs.google.com/spreadsheets/d/1Ty5X3LkY2unxTNGbkq36qDzuuYxQMsuywaTNyzL_pac/export?format=csv&gid=1746689811. Downloads /templates/google-sheets.csv and /templates/google-sheets.xlsx.) | {"format": "jsonstat2"} (SSB/Eurostat; filter the query to a single series), "transform"?: [{"op": "pct_change"|"diff"|"rolling_mean"|"rolling_sum"|"rolling_min"|"rolling_max"|"shift", "n": int} | {"op": "scale", "factor": num}], "keep"?: "all"|"last"}. "value_type": "text" stores labels (no transforms; see Data policy). Dates like 2026-09-30, 2026M08, 2026Q2/2026K2, 2026 or unix time are normalised to ISO period-end dates; date-times (2026-10-05T14:00+02:00) count on the day as written. A single JSON value with no timestamp (omit "date", or a date path the body doesn't have) is dated with the fetch's half hour: that is valid for a snapshot such as a price. A date that is present but not a date is an error. Do not invent a timestamp field. URLs and string bodies may use {today}, {yesterday}, {tomorrow}, {year}, {month}, {day} (UTC, filled in on every run) for sources that put the date in the URL. Private APIs: the URL, header values and string body may use {{secret:NAME}} for a secret the workspace owner stored (list_secrets shows the names and the one host each may be sent to); the value is filled in at fetch time and never shown, e.g. "headers": {"Authorization": "Bearer {{secret:STRIPE_KEY}}"}. A stored secret may be sent on GET, on a POST whose body is a GraphQL query, or on POST https://api.hubapi.com/crm/v3/objects/{deals,contacts,companies}/search only (HubSpot's read-only CRM search, JSON body; https://developers.hubspot.com/docs/api-reference/legacy/crm/search-the-crm). No other POST may carry a secret, and the secret is still sent only to the host it was saved for. If a needed secret is missing, ask the user to add it at /account under Data sources.

Read the latest annotations of a series.

List a dashboard's saved versions (newest first) with which one is live and a members-only preview URL for each.

List dashboards with their URLs.

List integration setup playbooks (Stripe, Shopify, Sheets, and the other sources tlly can read). A spreadsheet starts from the KPI template. Make a copy https://docs.google.com/spreadsheets/d/1Ty5X3LkY2unxTNGbkq36qDzuuYxQMsuywaTNyzL_pac/copy. Public CSV https://docs.google.com/spreadsheets/d/1Ty5X3LkY2unxTNGbkq36qDzuuYxQMsuywaTNyzL_pac/export?format=csv&gid=1746689811. Downloads /templates/google-sheets.csv and /templates/google-sheets.xlsx. Public: no API key required.

List scheduled fetch jobs with their last run status and any error message.

List the names of the workspace's stored secrets (API keys for private sources) and the host each may be sent to. Use one in a job's url, headers or body as {{secret:NAME}}.

List all series with their type (number or text), point counts and date ranges; text series also list their values.

Make a saved version of a dashboard the live one (also to roll back). The dashboard keeps its URL and visibility.

Write dated values to a series (upsert by date). A date is YYYY-MM-DD (midnight UTC), or a time such as 2026-09-30T14:30Z, floored to the half hour. Use when you compute data yourself. Values are numbers, or short text labels for a text series (e.g. a rating or stance; see Data policy). A series' first write fixes its type.

Give a dashboard a new public URL and disable the old one at once (e.g. if a link leaked). Spec and data are unchanged.

Run a scheduled fetch job immediately, store any new points, and return the run result.

Set or clear the workspace webhook URL for job-failure and dashboard status alerts (pass an empty url to clear). When events fire, tlly POSTs JSON to that URL (Slack and Discord incoming webhooks get a short text payload instead). Events include job_failed, status_changed (keep panel "id" fields in specs so alerts stay attached), and threshold_crossed. Email alerts work without a webhook. Caps: 12 emails and 60 webhooks per account per hour; the same panel is not alerted twice within an hour.

Set when one dashboard panel sends alert email and/or a webhook (colour change, threshold cross, or off). trigger: change (default), red, notgreen, above, below, or off; above/below need limit. notify: both, email, or webhook. panel is the 0-based index in the live spec. Saving can send immediately if a limit was just crossed (unless this panel alerted within the last hour). Later sends happen after scheduled fetches. Default change+both removes a custom rule. Does not change colour lines or save a dashboard version.

Configure when tlly sends stale-data alert email and webhooks because a series stopped updating. after: auto (from the job schedule and point spacing, at least two days), off, hours, or a window like 48h or 7d. Use dashboard_id for every series on a dashboard, series for one series, or both for one series on one dashboard. One email and one webhook when it goes quiet; another when a newer point arrives. Counts toward the hourly caps.

Dry-run a job spec: fetch, extract and transform, return a preview. Saves nothing. Job spec: {"series": name, "cron": "m h dom mon dow" (UTC, at most every 30 min), "request": {"url", "method"?: "GET"|"POST", "headers"?: {}, "body"?: obj|str, "pages"?: {"param": "page", "start"?: 0, "max"?: 1-10} (json with items: bump a query parameter until an empty items array) | {"cursor": "paging.next.after", "into": "after", "max"?: 1-10} (copy that JSON path from each response into the JSON body; stop when it is missing. A cursor that remains after max pages is an error, so a sum is never a partial total. HubSpot CRM search uses this. A json aggregate with no request.pages fails when the body still has another page, paging.next.after, has_more true, links.next, pageInfo.hasNextPage or another next cursor, instead of saving a partial total. Set pages. A response with no next marker is complete.)}, "extract": {"format": "json", "items"?: "path.to.array", "date"?: "field", "value": "field", "value_type"?: "number"|"text", "aggregate"?: "count"|"sum"|"mean"|"min"|"max" (records per date, e.g. hourly prices to a daily mean; needs items; all but count need value), "period"?: "D"|"M"|"Q", "where"?: [{"field": "path", "equals"|"not_equals": value}] (up to 5, with aggregate). Omit "date" with aggregate to snapshot every matching record as one point on the fetch's half hour. String numbers (HubSpot properties.amount) are summed.} | {"format": "json_columns", "dates": "path.to.array", "values": "path.to.array", "value_type"?} (two parallel arrays, e.g. Yahoo chart: timestamp + close) | {"format": "csv", "date"?: "COLUMN", "value": "COLUMN", "delimiter"?, "value_type"?} (omit date to stamp every row with the fetch's half hour; one value per half hour, so the last row wins. That does not sum the rows the way a JSON aggregate that omits date does. A history needs a Date column. Make a copy https://docs.google.com/spreadsheets/d/1Ty5X3LkY2unxTNGbkq36qDzuuYxQMsuywaTNyzL_pac/copy. Public CSV https://docs.google.com/spreadsheets/d/1Ty5X3LkY2unxTNGbkq36qDzuuYxQMsuywaTNyzL_pac/export?format=csv&gid=1746689811. Downloads /templates/google-sheets.csv and /templates/google-sheets.xlsx.) | {"format": "jsonstat2"} (SSB/Eurostat; filter the query to a single series), "transform"?: [{"op": "pct_change"|"diff"|"rolling_mean"|"rolling_sum"|"rolling_min"|"rolling_max"|"shift", "n": int} | {"op": "scale", "factor": num}], "keep"?: "all"|"last"}. "value_type": "text" stores labels (no transforms; see Data policy). Dates like 2026-09-30, 2026M08, 2026Q2/2026K2, 2026 or unix time are normalised to ISO period-end dates; date-times (2026-10-05T14:00+02:00) count on the day as written. A single JSON value with no timestamp (omit "date", or a date path the body doesn't have) is dated with the fetch's half hour: that is valid for a snapshot such as a price. A date that is present but not a date is an error. Do not invent a timestamp field. URLs and string bodies may use {today}, {yesterday}, {tomorrow}, {year}, {month}, {day} (UTC, filled in on every run) for sources that put the date in the URL. Private APIs: the URL, header values and string body may use {{secret:NAME}} for a secret the workspace owner stored (list_secrets shows the names and the one host each may be sent to); the value is filled in at fetch time and never shown, e.g. "headers": {"Authorization": "Bearer {{secret:STRIPE_KEY}}"}. A stored secret may be sent on GET, on a POST whose body is a GraphQL query, or on POST https://api.hubapi.com/crm/v3/objects/{deals,contacts,companies}/search only (HubSpot's read-only CRM search, JSON body; https://developers.hubspot.com/docs/api-reference/legacy/crm/search-the-crm). No other POST may carry a secret, and the secret is still sent only to the host it was saved for. If a needed secret is missing, ask the user to add it at /account under Data sources.

Replace a dashboard's spec (URL stays the same). Every update is saved as a numbered version. With "draft": true it is saved without going live: viewers keep seeing the live version, members can preview the draft at the returned preview_url, and publish_dashboard_version makes it live. Use drafts for larger changes and publish when the user confirms. Dashboard spec: {"title": str, "description"?: str, "range"?: "1m"|"3m"|"6m"|"ytd"|"1y"|"2y"|"5y"|"10y"|"max", "theme"?: "auto"|"light"|"dark" (default mode; auto follows the viewer's system and viewers can switch), "layout"?: "auto"|"overview"|"detail" (only the default for panels without "display": overview makes them tiles, detail makes them charts, auto makes them tiles above 8 panels), "visibility"?: "public"|"private" (default public: anyone with the link; private: only members of this account signed in to tlly.dev, so embeds and shared links stop working for others), "panels": [{"title"?: str, "series": "name" | ["a","b"], "type"?: "line"|"area"|"bar"|"stat"|"gauge"|"bargauge"|"pie"|"heatmap"|"table", "unit"?: str, "note"?: str, "stale_after_days"?: num, "thresholds"?: {"red_above"?, "yellow_above"?, "yellow_below"?, "red_below"?}, "range"?: (as above, overrides the dashboard's), "section"?: str (starts a new titled section at this panel), "views"?: bool (show the Level / Y/Y % / Index buttons; default: shown unless the unit is already a rate like %, y/y or pp), "display"?: "tile"|"chart" (tile: a compact number with a sparkline, click for the full chart; chart: a card with the full chart; mix them freely, e.g. tiles for the many supporting numbers and charts for the few that need their history), "headline"?: bool (show as a big number at the top of the dashboard, under a status strip; use for the 3-6 most important), "width"?: "half"|"full", "scale"?: "linear"|"log", "id"?: str (stable panel id, e.g. pn_abc123XYZ012; keep existing ids when editing so alerts stay attached), "right_axis"?: [series names] (line, area and bar charts: draw these series against a second y axis on the right, for series of different sizes or units; at least one stays on the left), "right_unit"?: str (unit of the right axis; "unit" is the left's), "left_label"?/"right_label"?: str (≤40 chars, a title written vertically along the left / right y axis, e.g. "Policy rate"; right_label needs right_axis), "view"?: "level"|"yoy"|"index" (initial view; viewers can switch), "source"?: str (shown as "Source: …"; defaults to the job's host), "labels"?: [display name per series], "status_series"?: name (a red/yellow/green text series that sets the panel colour), "compare"?: "prev"|"3m"|"1y" (what the change is measured against), "period"?: "D"|"M"|"Q" (show dates as 2026-06 or 2026Q2), "value_names"?: {"green": "Hold", …} (display names for text values)}], "badge"?: "worst" (default: the header badge shows the worst panel) | "off" | panel index (0-based; the badge follows that panel), "shade"?: [{"from": date, "to": date, "label"?: str}] (up to 20 shaded periods on every chart)}. "type": "table" shows one row per series (up to 20): label, latest value, date and a status dot; its "status_series" is a list with one entry per row ("" for none). "area" is a line chart filled underneath; "gauge" is a dial of the latest value with the threshold zones, and "bargauge" a horizontal bar per series (up to 20, coloured by thresholds), both scaled by "min"?/"max"?: num (default 0 to a bit above the largest value or threshold); "pie" is a donut of the series' latest values (shares of a whole, up to 20); "heatmap" is a calendar of the first series' daily values (good for activity per day). Status colour is computed from the latest value of the panel's first series; a series of red/yellow/green labels (e.g. a derived status) colours itself. Notes may use {latest}, {date}, {period} and {change}, filled in from the panel's latest value. Annotations on a panel's series appear as markers. Viewers can pick a time range (1M to Max, year to date, or a custom from-to period), zoom, switch views and download CSV; ?embed=1 hides the page chrome and &panel=N shows one panel. Text limits: title ≤120, description ≤500, panel title ≤120, unit ≤20, note ≤280, section and source ≤80 characters, single line, plain text.

Replace a job's spec, or pause it with enabled=false. Job spec: {"series": name, "cron": "m h dom mon dow" (UTC, at most every 30 min), "request": {"url", "method"?: "GET"|"POST", "headers"?: {}, "body"?: obj|str, "pages"?: {"param": "page", "start"?: 0, "max"?: 1-10} (json with items: bump a query parameter until an empty items array) | {"cursor": "paging.next.after", "into": "after", "max"?: 1-10} (copy that JSON path from each response into the JSON body; stop when it is missing. A cursor that remains after max pages is an error, so a sum is never a partial total. HubSpot CRM search uses this. A json aggregate with no request.pages fails when the body still has another page, paging.next.after, has_more true, links.next, pageInfo.hasNextPage or another next cursor, instead of saving a partial total. Set pages. A response with no next marker is complete.)}, "extract": {"format": "json", "items"?: "path.to.array", "date"?: "field", "value": "field", "value_type"?: "number"|"text", "aggregate"?: "count"|"sum"|"mean"|"min"|"max" (records per date, e.g. hourly prices to a daily mean; needs items; all but count need value), "period"?: "D"|"M"|"Q", "where"?: [{"field": "path", "equals"|"not_equals": value}] (up to 5, with aggregate). Omit "date" with aggregate to snapshot every matching record as one point on the fetch's half hour. String numbers (HubSpot properties.amount) are summed.} | {"format": "json_columns", "dates": "path.to.array", "values": "path.to.array", "value_type"?} (two parallel arrays, e.g. Yahoo chart: timestamp + close) | {"format": "csv", "date"?: "COLUMN", "value": "COLUMN", "delimiter"?, "value_type"?} (omit date to stamp every row with the fetch's half hour; one value per half hour, so the last row wins. That does not sum the rows the way a JSON aggregate that omits date does. A history needs a Date column. Make a copy https://docs.google.com/spreadsheets/d/1Ty5X3LkY2unxTNGbkq36qDzuuYxQMsuywaTNyzL_pac/copy. Public CSV https://docs.google.com/spreadsheets/d/1Ty5X3LkY2unxTNGbkq36qDzuuYxQMsuywaTNyzL_pac/export?format=csv&gid=1746689811. Downloads /templates/google-sheets.csv and /templates/google-sheets.xlsx.) | {"format": "jsonstat2"} (SSB/Eurostat; filter the query to a single series), "transform"?: [{"op": "pct_change"|"diff"|"rolling_mean"|"rolling_sum"|"rolling_min"|"rolling_max"|"shift", "n": int} | {"op": "scale", "factor": num}], "keep"?: "all"|"last"}. "value_type": "text" stores labels (no transforms; see Data policy). Dates like 2026-09-30, 2026M08, 2026Q2/2026K2, 2026 or unix time are normalised to ISO period-end dates; date-times (2026-10-05T14:00+02:00) count on the day as written. A single JSON value with no timestamp (omit "date", or a date path the body doesn't have) is dated with the fetch's half hour: that is valid for a snapshot such as a price. A date that is present but not a date is an error. Do not invent a timestamp field. URLs and string bodies may use {today}, {yesterday}, {tomorrow}, {year}, {month}, {day} (UTC, filled in on every run) for sources that put the date in the URL. Private APIs: the URL, header values and string body may use {{secret:NAME}} for a secret the workspace owner stored (list_secrets shows the names and the one host each may be sent to); the value is filled in at fetch time and never shown, e.g. "headers": {"Authorization": "Bearer {{secret:STRIPE_KEY}}"}. A stored secret may be sent on GET, on a POST whose body is a GraphQL query, or on POST https://api.hubapi.com/crm/v3/objects/{deals,contacts,companies}/search only (HubSpot's read-only CRM search, JSON body; https://developers.hubspot.com/docs/api-reference/legacy/crm/search-the-crm). No other POST may carry a secret, and the secret is still sent only to the host it was saved for. If a needed secret is missing, ask the user to add it at /account under Data sources.