io.github.tickerbot/mcp-server

A named boolean predicate you can reference anywhere a built-in signal goes.

Create a named ticker list owned by your account.

Canonical create: a webhook is a trigger plus a delivery. Trigger shapes: scan {type:"scan", q, universe?}; ticker {type:"ticker", ticker, condition}; signal {type:"signal", signal, ticker?, universe?, condition?}; event {type:"event", kinds, tickers?, universe?, event_q?}. The subscribe tools are flat sugar over exactly this. Webhooks need a paid plan (Free has no webhook slots).

Delete one of your custom signals. Refused by default if another of your signals references it.

Permanently delete one of your universes. Webhooks that reference the deleted universe fail on their next eval — clean those up first.

Delete a webhook subscription, and its delivery history with it.

Re-enable a disabled webhook and start it clean. Clears match-state, so the next eval treats every currently-matching ticker as new.

OHLCV bars from 1-second through monthly. The prices underneath the table.

Returns an ETF's constituents and their weights, heaviest first. When the ticker is not an ETF, `is_etf` is false and `holdings` is empty; `is_etf: true` with zero holdings means a real ETF whose holdings are not ingested yet. The reverse lookup ("which ETFs hold NVDA") is a scan filter on `etf_holders`, not this tool.

Returns an ETF's sector weights, heaviest first. Always complete, since sector breakdowns are small. When the ticker is not an ETF, `is_etf` is false and `sectors` is empty.

Any signals for any tickers on one shared time grid — up to 50 tickers by 25 columns per call. One flat row per ticker per interval step, cursor-paged backward. `transitions_only: true` with boolean signals returns only the rows where a boolean flipped.

The state of a signal is the set of tickers matching it right now, or with `asof`, as of any moment. One name, the whole market, one call. Booleans need no `condition`; every other type needs one, typed like the signal: ">70" (numeric), "<YYYY-MM-DDTHH:MM:SSZ" (timestamp), ">=YYYY-MM-DD" (date), "=ETF" (string). Sorted by signal value desc for non-booleans.

The full ticker row, every signal on the schema page, for one symbol or a comma list of up to 50. Right now, or with `asof`, as of any past date. Pass a comma list of up to 50 symbols for a batch (`data` keyed by symbol plus `not_found`). Crypto is the X-prefixed pair (X:BTCUSD) — bare BTC/ETH are US-listed ETFs.

For one ticker, what we hold and how far back — so an empty result is never ambiguous. Ask this before treating a gap in bars or series as an outage. An absent `minute` span is not a gap — sub-hour bars load on the first request for a symbol.

Returns the universe doc. Use `top_10`/`top_100` to fetch a system universe; any other slug must be one your account owns.

The current state of one webhook subscription.

One timeline across every ticker: earnings, dividends, splits, insider filings, analyst actions, plus opt-in signal firings and news. Requires at least one bound: a ticker scope (ticker/tickers/universe), a time window (from/to), or firm/action — `q` alone is not a bound. `firm`/`action` match case-insensitively; a `q` payload match is case-sensitive. `join: state` attaches the ticker state as of each event.

Every signal you can name in a query — the built-in signals and your own custom signals, in one catalog. Use to discover the signal names and `q` vocabulary before composing a scan; custom signals appear with `kind: custom`.

Every symbol we track, active or delisted, as one identity row each. Use `/v2/tickers/{ticker}` for the full row.

Every universe you can reference: your own named ticker lists and the built-in ones. `owner: system` lists the built-in universes (top_10, top_100); `all` lists both. Use a slug as `universe` on scan, signal, and subscribe tools.

Recent deliveries for one webhook — what was sent, and what came back.

Every webhook subscription on this account, newest first.

Every ticker matching a SQL WHERE clause. Right now, or with `asof`, as of any past date. The `q` grammar is a flat SQL WHERE over signal names: AND/OR/NOT, comparisons, numeric and string literals, custom signals by name. No JOIN or subqueries. With `group_by` the result is rollup rows, not tickers. Example: `gap_up AND market_cap < 2000000000 AND NOT earnings_this_week`.

SQL query over the news archive. Article rows, or rollups when you group them. Filter to a ticker with the `ticker` param, or in `q` via the auto-unnest alias `tk = 'NVDA'`. `search` is full-text over title and summary. Rollups with group_by/having return `truncated: true` instead of paging.

Push new events: we POST your endpoint when events of the kinds you chose land in the archives. Webhooks need a paid plan (Free has no webhook slots). `q` filters the ticker STATE; `event_q` filters the EVENT payload in the /v2/events grammar. Latency is the ingest cadence (analyst ≤1h, corporate kinds daily), not sub-minute.

Push a whole query: we POST your endpoint every time the match set changes. Webhooks need a paid plan (Free has no webhook slots). Use for "alert me when this happens" requests.

Push one signal: we POST your endpoint whenever any ticker starts matching it. Webhooks need a paid plan (Free has no webhook slots). Omit `ticker` to watch the whole universe.