agentleFS
MendurimMendurim/strikegrok-trading-desk/skills/strike-advanced/SKILL.md

strike-advanced

Skill0 starsChanged 2 months agoMITView on GitHub

Strike Finance execution features beyond a single order - TWAP strategies, batch and batch-replace orders, cancel-all, isolated-margin adjustment, post-only and price-protect flags, and trailing stops - with the approval each needs and the ones this desk refuses. Use when a ticket calls for a TWAP or a multi-order send, when adjusting isolated margin, or when deciding whether a feature is in scope. Write paths - Execution Trader only.

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read the response per order.

    line 61
  2. Very specific:

    Never as a reflex, never to "clean up", and never before checking GET /v2/positions to see what the cancelled orders were protecting.

    line 72
  3. Very specific:

    No vault_id: Every endpoint accepts one, to trade on behalf of a vault as its leader.

    line 95

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Strike advanced

Everything here writes, and everything here is a ticket decision, not an execution flourish. Read strike-orders first.

The rule that governs this whole skill: a feature the Risk Manager did not size for is a feature the desk does not use. Each of these changes the shape of the risk, so each belongs on the ticket, in the PASS, and in the user's approval.

1. TWAP

POST /v2/algo/twap slices an order over time. For a size the book cannot absorb in one go, this is the honest way to execute it - and the Market Analyst's depth read is what says whether it is needed.

{
  "symbol": "ADA-USD",
  "side": "BUY",
  "total_size": "250000",
  "duration_sec": 1800,
  "limit_price": "0.2150",
  "reduce_only": false,
  "randomize": true
}
  • side is uppercase here (BUY / SELL), unlike the lowercase buy / sell on /v2/order. An easy and silent mistake.
  • limit_price bounds the slices. Set it. A TWAP without one is a standing instruction to pay whatever the book asks for half an hour.
  • randomize jitters slice timing.
python3 scripts/strike_request.py POST /v2/algo/twap --body-file twap.json
python3 scripts/strike_request.py GET /v2/algo/twap                    # list running strategies
python3 scripts/strike_request.py GET /v2/algo/twap/<id>               # one strategy
python3 scripts/strike_request.py DELETE /v2/algo/twap/<id>            # cancel it

A running TWAP is an open commitment. It keeps sending while nobody is watching, which is exactly the thing the one-writer rule exists to bound. So:

  • A TWAP ticket states total size, duration, limit price and the protection that will be attached when it completes.
  • The desk holds no other position-opening order on that symbol while it runs.
  • It is monitored: GET /v2/algo/twap/<id> on a schedule, and the journal records each check.
  • Protection cannot attach until there is a position, so the Execution Trader places the stop as soon as the first slice fills, sized to the position so far, and resizes it (place new, confirm, cancel old) as the TWAP builds. A half-built TWAP position with no stop is an unprotected position.
  • Cancelling a TWAP is its own approved action, except when protection has failed - then it is an incident and the desk cancels first and reports immediately.

2. Batch orders

POST /v2/orders/batch takes an orders array of the same objects /v2/order accepts. POST /v2/order/replace-batch replaces several at once.

Legitimate uses on this desk: laddering an entry the Risk Manager sized as one position across several price levels, and resizing several protective stops together after a partial fill.

Not legitimate: bundling unrelated tickets. One approval, one send still holds - a batch is one send of one ticket's orders, never a convenient way to execute two decisions at once. Every order in the batch carries its own client_order_id derived from the same ticket id.

Read the response per order. A batch can partially succeed, and a batch where three of five rested is a state the desk must reconcile order by order, not summarise.

3. Cancel-all

DELETE /v2/order/cancel-all, optionally scoped by symbol.

This is a blunt instrument: unscoped, it removes protective stops on positions that are still open, turning a tidy-up into a naked book. The desk uses it only:

  • scoped to a symbol, when flattening that symbol and the position is already confirmed closed; or
  • under an incident playbook the user has pre-authorised in desk.md.

Never as a reflex, never to "clean up", and never before checking GET /v2/positions to see what the cancelled orders were protecting.

4. Isolated margin

POST /v2/marginMode switches a symbol between cross and isolated; Strike rejects the change while a position is open on that symbol.

POST /v2/isoMargin adds or removes margin on an isolated position (symbol, amount, modify_type).

Adding margin moves the liquidation price away and is risk-reducing; removing it moves the liquidation price closer and is increasing risk, so it needs the same approval as opening a position. Re-read GET /v2/positions afterwards and report the new liquidation distance in price and percent. On an isolated position it is the position's own margin, not account equity, that stands between it and liquidation.

5. Order flags worth knowing

FlagEffectDesk position
post_onlyrejected rather than crossing; guarantees makerfine on a resting entry the ticket wants as maker
price_protectexchange-side protection against triggering on a bad printon for triggers unless the ticket says otherwise
close_positiontrigger closes the whole position, whatever the size field sayspreferred for a flattening stop; survives partial fills and adds
working_typemark_price (default) or contract_price for trigger evaluationmark_price, because that is what liquidation settles against
slippagebound on a market order, decimal fractionalways set on a market order
callback_ratetrailing distance, "0.1" to "5" percentonly when the ticket specifies a trailing stop

6. What this desk does not do

  • No vault_id. Every endpoint accepts one, to trade on behalf of a vault as its leader. The desk trades the user's own account only, and never passes the field.
  • No deposits, withdrawals, bridging or transfers. The API wallet cannot do them and the desk would not if it could; those happen in the Strike app, with the user.
  • No unattended sends. Routines and watches alert and draft. A TWAP is the one thing on the desk that keeps executing after the send, which is why it is monitored and bounded rather than treated as fire-and-forget.
  • No dead-man's switch. Strike has no such endpoint. Do not improvise one, and do not tell the user the desk has that protection.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.