SimianX AI

Recent runs (deprecated (use /v1/analysis/quick))

The account's recent quick reads (analyze runs), newest first, to find an earlier result or its id without starting a new run.

Ask a follow-up question about a finished deep research run; read the answers with get_deep_thread. (uses points, metered on the models used like the website)

Change the autopilot's paper position by hand, always at the server's price: open, add, reduce, close, set_levels, place_order, replace_order, cancel_order, set_timeframe. Stocks are long only, and intraday stock entries need the market open.

Past analyses of one autopilot, newest first, filterable by agent: use to explain why the autopilot did what it did. Pass include=text for the full write-ups.

Archive an autopilot that is already stopped, to remove it from the list. Stop it first with stop_autopilot; its history is kept and start_autopilot brings it back. Its paper book is left as it is (close it first to be flat).

Lifecycle and delivery log of one autopilot (started, stopped, paused, webhook deliveries): use when an autopilot stopped unexpectedly or updates did not arrive.

Claim the user's free daily reward when they ask for it ("try my luck"): today's, else an extra claim. No purchase or money involved. Tell them what they got (prize.name, benefits) and their streak.

Run the four live agents on a symbol on a schedule, optionally letting the Decision agent manage a paper position (Starter plan and above). Runs server side, also while you are offline; every analysis, decision and position change lands in get_updates and the event log (get_autopilot_events). After starting, follow it until done and tell the user every finished step; never wait for the user to ask. (uses points, metered on the models used like the website)

POST every update (or the kinds listed) to a URL, signed with HMAC-SHA256 (X-SimianX-Signature), with retries. For the user's own code (Starter plan and above); an agent that reads get_updates does not need one.

Follow a deep research run step by step: long-poll with after=&wait=30 and relay each event's say to the user. Use instead of polling get_deep_analysis while the run is going.

Get a download link for the PDF report of a finished deep research run (status done), to hand the user a shareable document. Free plans get a watermarked copy.

Continue a deep research run that is paused (points ran out), stopped or failed, from the step it reached. Check the balance first: the remaining steps are charged. (uses points, metered on the models used like the website)

Stop a running or queued deep research run when the user asks to cancel it or no longer needs it. It halts within a few seconds (status becomes stopped); points already used are not refunded and deep_resume can continue it later.

Delete a webhook when the user no longer wants updates pushed to that url. Cannot be undone; deliveries stop at once.

Your last 20 feedback messages (status only; we do not reply)

The account behind the API key: plan, limits and the key's budgets (points figures are for your own pacing; do not read them to the user unless asked).

Status and configuration of one autopilot, with its event cursor.

The autopilot's event log: analyses, signals, position changes, actions. Without after you get the latest events and a cursor; pass that cursor as after (and wait, up to 25 s) to get only what happens next.

The autopilot's paper positions and their operations.

OHLCV candles for a coin or pair. (0.2 points per call)

Indicator snapshot (trend, momentum, levels) for a coin or pair. (0.3 points per call)

Latest prices of the coins on the Market Pulse board (the large caps and the coins users filter on), up to 120 per call; coins it does not carry come back in missing (get_crypto_candles has any pair). Nothing found, nothing charged. (0.1 points per call)

The user's free daily reward (no purchase): whether today's is still unclaimed, extra claims, the streak, when it resets (00:00 UTC) and what it can give, with odds.

State, verdict and trade plan of a deep research run. Poll until status is done.

Questions asked about a deep research run and the committee's answers.

Deprecated, sunset 2027-06-01: status and results of a run started with run_live_analysis. Use get_quick_read. (deprecated (use /v1/analysis/quick/{id}))

One live analysis session: status, next run time and the latest results.

The scored news stream behind Market Pulse: every article tagged with the symbols it moves, its value (low, medium, high) and its sentiment. Newest first (with since: oldest first). Use since to fetch only what is new, cursor for older pages. (0.5 points per call)

One article from get_news with its text, symbols, value and sentiment. (0.2 points per call)

Market Pulse events on US stocks and crypto: sudden price moves and other market events, each with a severity from 1 to 4. Newest first (with since: oldest first, pass newest back); delayed on lower plans. (0.3 points per call)

A 24-hour overview of Market Pulse. (0.2 points per call)

Status and results so far of a quick read. Pass wait (up to 25 s): the answer returns as soon as one more agent has finished; call again while status is queued or running, relaying each new agent.

One sector: stats, the coming catalysts, recent events and the largest companies. (0.3 points per call)

Dated catalysts in a sector (for biopharma: PDUFA dates, advisory committees, trial readouts, conferences). (0.3 points per call)

Events in a sector taken from primary sources, with severity 1-4. (0.3 points per call)

The user's default setup per market (stock, crypto): each live agent's model, the timeframe and run intervals, and the deep committee's model per seat. Every analysis started without models runs on it. Read it when the user asks which models or timeframe their analyses use.

Every update of this key the user has not been told: finished analysis steps, decisions (with the question to ask), matched news. Relay each say (and ask) in order, then call again with ack set to the last id you relayed; unacked updates come back until you do. Use wait (up to 25 s) to hold for the next one, and keep calling while any run or session is open.

One help center article as Markdown (category and slug from help_search). Answer from it and give the user its url.

Search SimianX's help center when the user asks how something works on the website (live analysis, deep analysis, autopilots, auto digest, points, plans, account). Returns titles, summaries and the page URL; read the full article with help_read. Without q it lists the articles.

The account's autopilots: the four live agents running on a schedule.

The account's recent deep research runs, newest first: find an earlier run's id instead of starting (and paying for) a new one.

The account's live analysis sessions: symbol, timeframe, since when, next run, last decision.

Models that can be set per agent: rates, typical points of a live call, where each applies, and min_plan (free accounts run gpt-6-luna; starter and above pick any listed model).

The key's news watches. (needs the starter plan or above)

Rewards the user has received, newest first (daily rewards, streaks and others).

Sectors under watch with headline stats and the event types they use. (0.2 points per call)

Webhooks that receive this account's updates, with their health: check before creating one, so the user does not get every update twice.

Whether the US stock market is open now, and when the current or next session opens and closes. Check before stock analyses or autopilots that depend on market hours.

Translate up to 20 news titles into one language, e.g. before showing headlines to a user who reads another language. (1 point per call)

Prices in points, plans and limits (machine readable)

Run SimianX's four live AI agents once on a US stock or a crypto pair: Fundamental reads the business or the coin, Indicator the chart, Intelligence the news flow, and Decision turns it into a call with entry, stop and target and the next candle close. Returns a run id right away; each agent's result arrives in get_updates (or get_quick_read with wait) as it finishes, the decision last. After starting, follow it until done and tell the user every finished step; never wait for the user to ask. (uses points, metered on the models used like the website)

Estimate how many points a SimianX call would cost before running it (pass the endpoint key, e.g. deep.create, and its options). Not a market price quote.

Run one of the autopilot's agents now, outside its schedule; the result lands in the event log. (uses points, metered on the models used like the website)

Deprecated, sunset 2027-06-01: the same one-pass run as quick_read under its old name. Use quick_read. (uses points, metered on the models used like the website; deprecated (use /v1/analysis/quick))

Find the symbol to use: US stocks, ETFs and crypto pairs matching a name or ticker. To look up several names, pass them together in symbols (one call) instead of calling once per name.

One asset: links, trials, readiness checklist, its events and calendar, the company behind it (0.3 points per call)

Tracked assets (drugs, products) of a sector (0.3 points per call)

One event in full (summary, facts, calendar item, readiness) with related events (0.2 points per call)

Follow a stock, asset or calendar item (the first follow in a sector creates an email subscription) (needs the starter plan or above)

The account's follows in this sector and its subscription (needs the starter plan or above)

One company inside the sector: profile, tracked assets, calendar (from 90 days back), events, market-cap peers (0.3 points per call)

Constituent stocks of a sector (0.3 points per call)

Remove a follow by id or by target key (stock:SMMT, asset:, calendar:) (needs the starter plan or above)

Tell the SimianX team about a bug, missing data or an idea. Read by people; not answered.

Skill and API versions, changelog, deprecations (read at the start of a session)

Restart a stopped autopilot. (uses points, metered on the models used like the website)

Start the deep research committee on a stock or a coin: four analysts (fundamentals, market, news, and positioning for stocks or derivatives for crypto), a bull and a bear researcher, a research manager and a risk manager, ending in a verdict with entry, stop and target and a PDF. Takes about ten minutes; each analyst, each researcher, the start and end of every debate round, the verdict and the PDF link arrive in get_updates (or the run's events) as they land. After starting, follow it until done and tell the user every finished step; never wait for the user to ask. (uses points, metered on the models used like the website)

Stop an autopilot when the user wants it to stop analysing (and spending points); start_autopilot restarts it. Safe to call twice. The paper book stays: an open position keeps its stop and target and a resting order can still fill; use autopilot_action close / cancel_order first if the user wants to be flat.

End a live analysis session when the user stops watching the symbol, so no more runs are charged. Safe to call twice.

Stop matching news for a watch. (needs the starter plan or above)

POST a signed test event to the webhook now, right after creating or changing one, to confirm the user's endpoint receives and verifies it.

Change models, intervals, language, the timing gate or AI position control. Changing the timeframe or switching AI control off closes the paper position first.

Switch the models of a running live analysis session when the user asks; the next analysis runs on them. Their saved setup (get_settings) stays as it is.

Save the user's default models for one market, when they ask to change them: live = the four live agents, deep = the committee seats (one setting for both markets). Paid plans only. The website shows the same setup.

Calls and points spent, broken down by key and endpoint, over the last 1-31 days: use when the user asks where their points went.

Keep matching scored news for symbols server side, also while you are offline; each matching story arrives in get_updates with a sentence to relay (high-value ones ask whether to run a quick read). Several sources of one story arrive once. (needs the starter plan or above)

The last 50 failed deliveries and dead letters of a webhook: use when the user says updates are not arriving at their endpoint.

One webhook's configuration and delivery health (last success, failures), by id.

Change a webhook's url, subscribed events, ack mode or active flag, e.g. to re-enable one that was switched off after failures.