agentsclimarketplace

Trading orders

Skill alpacahq/alpaca-skills/skills/broker-api/trading-orders

Place and manage orders on behalf of accounts via the Alpaca Broker API — order creation (qty vs notional, fractional shares, order types/TIF/classes), order status lifecycle, replace/cancel, positions, and trading-account buying power. Use when building trading, recurring-invest, or portfolio flows on Alpaca in any language.From its SKILL.md

Install
npx -y skills add alpacahq/alpaca-skills --skill trading-orders

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

SKILL.md

9.2 KB, ~2.5k tokens by cl100k_base, as published. Nobody here has run it

Alpaca Broker API — Trading on Behalf of Accounts

Place, modify, cancel, and track orders for an end-user account, and read positions & buying power. The defining feature of Broker API trading: account_id is in the path — you act for a user account, not your own.

Read alpaca-broker-integration first. Broker API + HTTP Basic auth. (The standalone Trading API uses /v2/orders with no account in the path; everything else here transfers.)

Reference

  • Guides: https://docs.alpaca.markets/docs/orders-at-alpaca, https://docs.alpaca.markets/docs/fractional-trading
  • API ref: https://docs.alpaca.markets/reference/postorder
  • Live schema: alpaca-docs MCP → get-endpoint title "Broker API" path /v1/trading/accounts/{account_id}/orders

1. Endpoints

MethodPathPurpose
POST/v1/trading/accounts/{id}/ordersCreate order
GET/v1/trading/accounts/{id}/ordersList orders (filter by status, symbols, after…)
GET/v1/trading/accounts/{id}/orders/{order_id}Get order by ID
GET/v1/trading/accounts/{id}/orders:by_client_order_id?client_order_id=…Get by your client ID
PATCH/v1/trading/accounts/{id}/orders/{order_id}Replace (modify) order
DELETE/v1/trading/accounts/{id}/orders/{order_id}Cancel one order (204)
DELETE/v1/trading/accounts/{id}/ordersCancel all (207 Multi-Status)
POST/v1/trading/accounts/{id}/orders/estimationCost-estimate an order
GET / DELETE/v1/trading/accounts/{id}/positions[/{symbol_or_asset_id}]List / close positions
GET/v1/trading/accounts/{id}/accountTrading-account details (buying power etc.)

2. Create-order request

Schema-required: type and time_in_force. Conditionally required: symbol, side, and exactly one of qty/notional.

// notional market buy (dollar-based, fractional)
{ "symbol": "AAPL", "notional": "25.00", "side": "buy", "type": "market", "time_in_force": "day",
  "client_order_id": "your-own-uuid" }

// limit qty sell
{ "symbol": "AAPL", "qty": "3", "side": "sell", "type": "limit", "limit_price": "190.00", "time_in_force": "gtc" }
FieldValues / notes
symbolrequired (except mleg multi-leg options)
qtydecimal string, up to 9 dp. Fractional only for market+day
notionaldecimal string, up to 9 dp. Mutually exclusive with qty
sidebuy, sell (plus advanced: sell_short, …)
typemarket, limit, stop, stop_limit, trailing_stop
time_in_forceday, gtc, opg, cls, ioc, fok
limit_price / stop_pricerequired for limit/stop variants
trail_price / trail_percentone required for trailing_stop
extended_hoursbool; only with type=limit and TIF day/gtc
client_order_id≤128 chars; your idempotency key (auto-generated if omitted)
order_classsimple (default), bracket, oco, oto, mleg
take_profit / stop_loss{limit_price} / {stop_price, limit_price?} for bracket/oco/oto
position_intentbuy_to_open, sell_to_close, …

qty XOR notional (verbatim rule): pass one or the other — supplying both → 400. In the response, whichever you didn't use comes back null.

3. Fractional / notional rules

  • On by default for all accounts (live + paper).
  • Asset must have fractionable: true (check the Assets API — see alpaca-broker-market-data), else requested asset is not fractionable.
  • TIF must be day for fractional/notional.
  • Notional is limited to market and limit (day); only limit for extended hours. Fractional qty additionally allows stop/stop_limit per the guide.
  • No shorting fractional — all fractional sells are marked long.
  • Precision: up to 9 decimal places for both qty and notional.

4. Order status lifecycle

OrderStatus (the order object's status): new, partially_filled, filled, done_for_day, canceled, expired, replaced, pending_cancel, pending_replace, accepted, pending_new, accepted_for_bidding, stopped, rejected, suspended, calculated.

Order status ≠ trade-event event. The order object's status is the enum above. The SSE trade-update stream reports a richer event enum that adds operational events not present as a status — including held (multi-leg secondary legs awaiting trigger), trade_bust, trade_correct, restated, order_cancel_rejected, order_replace_rejected. So held exists as a trade event but never as an order status. See alpaca-broker-sse-events.

Terminal: filled, canceled, expired, rejected (and replaced for the original order). Everything else is in-flight.

Early-state distinctions (these trip people up):

  • accepted — received by Alpaca, not yet routed to a venue (common outside market hours).
  • new — received and routed to exchanges; the usual initial live state.
  • pending_new — routed but not yet accepted for execution (rare).

So the typical opening sequence is accepted → pending_new → new, then fills. Lesson: treat new/accepted/pending_new as "exists but not done." Persist the order on submit, then update on fill/cancel/reject events — don't block the user waiting for a terminal state synchronously.

5. Positions & trading account

Position key fields: symbol, asset_id, qty, qty_available (free of open orders), side (long/short), avg_entry_price, market_value, cost_basis, unrealized_pl, unrealized_plpc, current_price, change_today.

TradeAccount key fields:

  • buying_power (with margin multiplier 1–4), cash, cash_withdrawable, equity, last_equity.
  • Blockers: trading_blocked, account_blocked, transfers_blocked, trade_suspended_by_user.
  • multiplier, regt_buying_power, non_marginable_buying_power, long_market_value, initial_margin, maintenance_margin, sma.

Lesson — check buying power before notional orders. For a "spend $X" UX, read buying_power/cash first and reject/notify on insufficient funds, rather than letting Alpaca reject the order. (Cache it per account within a batch run to avoid re-fetching.)

PDT/day-trade fields are deprecated (since 2026-04-27, sunset 2026-07-06) following FINRA's intraday-margin rule change: daytrade_count, pattern_day_trader, daytrading_buying_power, bod_dtbp, plus config dtbp_check/pdt_check. They still exist in the schema today but stop relying on them.

6. Documented gotchas

  • Wash-trade rejection (403): if a user's two orders could self-cross (opposite sides, crossable prices), Alpaca rejects. Opposing market/stop pairs are always rejected; opposing limits rejected when buy-limit ≥ sell-limit. Use bracket/oco/trailing_stop for simultaneous take-profit + stop-loss — they're exempt.
  • Bracket constraints: requires both take_profit.limit_price and stop_loss.stop_price; TP must be above SL for a buy; no extended hours; TIF day/gtc; child legs activate only after the entry fully fills; canceling one cancels the group.
  • Notional orders can't be replaced — cancel and resubmit (IPO-class notional is the exception). Fractional qty can't be changed on replace ("full shares only").
  • Replace ≠ guaranteed: a 200 from PATCH can still be rejected if the original fills first; watch the trade-updates stream. Can't replace while accepted/pending_new/pending_cancel/pending_replace.
  • Cancel semantics: single cancel → 204, or 422 if no longer cancelable; cancel-all → 207 per-order results; close-all positions → 207. Close-single accepts mutually-exclusive qty or percentage.

7. Idempotency & recurring-invest lessons

  • Always set client_order_id from your own transaction record. It's your dedup key and lets you look the order up (orders:by_client_order_id) if the create response is lost. Note it dedups lookup, not necessarily replay — combine it with a local "already-submitted?" guard.
  • Recurring/scheduled buys (lesson): the robust pattern is — fetch pending invest instructions from your DB → check buying power → place a notional market/day order per instruction → record the returned order → mark the instruction done only after a successful create. On insufficient funds, cancel the instruction and notify, don't silently skip. Schedule the batch shortly before market open and respect the market clock (alpaca-broker-market-data).
  • Track fills via the trade events SSE stream, not by polling each order — see alpaca-broker-sse-events.

Related skills: prices/assets/clock → alpaca-broker-market-data; fills in real time → alpaca-broker-sse-events; rate limits on bulk placement → alpaca-broker-rate-limits-resilience; money formatting → alpaca-broker-money-precision.

What ships with it: 1 file

1.3 KB alongside SKILL.md

Gives 0 of the 12 instructions most finance skills give in ~2.5k tokens

Counted across 469 of the 469 authors here whose files we hold, read 2026-08-07

  • Extract date vendor amount and descriptionin 15 of 469, across 3 files
  • Scan folder for invoice filesin 14 of 469, across 2 files
  • Rename files to standard formatin 14 of 469, across 2 files
  • Show organization plan before movingin 14 of 469, across 2 files
  • Generate summary CSVin 14 of 469, across 2 files
  • Organize files by categoryin 13 of 469, across 1 file
  • Preserve original filesin 13 of 469, across 1 file
  • Flag files missing critical infoin 13 of 469, across 1 file
  • Produce the requested output filein 9 of 469, across 4 files
  • Build best, base, and worst case scenariosin 9 of 469, across 5 files
  • Implement backoff if rate limit errors occurin 8 of 469, across 3 files
  • Determine the weighted average cost of capitalin 8 of 469, across 4 files

Said here and by no other author read

  • put account_id in the request path
  • use basic auth
  • include type and time_in_force in every order
  • supply exactly one of qty or notional
  • check buying power before placing notional orders
  • set client_order_id for idempotency

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.