agentleFS
galleonlabsgalleonlabs/hypergrok-trading-desk/skills/desk-post-trade-review/SKILL.md

desk-post-trade-review

Skill530 starsChanged 5 days agoMITView on GitHub

The Trade Reviewer's procedure for journaling desk activity and reviewing trades from the exchange record - process graded separately from outcome, execution costs measured, one repeatable finding per review, plus the weekly desk review. Use after any send, when a trade closes, on the weekly routine, or when the user asks "how did that go".

2 other projects use a version of this, 1 of them edited.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 1 is distinctive, and 1 is 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.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Write the review under ## review in the proposal file and send the block (format in agents/trade-reviewer.md) by DM to the Desk Lead and …

    line 63
  2. Distinctive:

    No recommendations about what to trade.

    line 76

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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

Journal and post-trade review

The journal is the desk's memory; the review is how the desk learns. Both come from the exchange record first and chat second, and both keep process and outcome apart.

1. Journal

/workspace/trading-desk/journal/YYYY-MM-DD.md, one file per active day, appended in time order. Entry types:

14:02 UTC  HG-20260816-01 opened   ETH-PERP long idea (user); status evidence
14:12 UTC  HG-20260816-01 risk     PASS 0.4827 ETH, stop 2,900, 0.5% risk stressed (risk-limits v3)
14:29 UTC  HG-20260816-01 approval "approve HG-20260816-01" (user)
14:31 UTC  HG-20260816-01 sent     buy 0.4827 @ 3,000 Gtc + sl 2,900 (normalTpsl); cloid 0x9f3e...; oid 1839201122, sl waitingForFill
16:05 UTC  HG-20260816-01 fill     0.4827 @ 3,000.0 maker, fee $0.29
09:12 UTC  HG-20260816-01 closed   tp 3,089.6, fee $1.44; position flat; sl 2,900 cancelled 09:13
10:00 UTC  limits                  risk-limits v3 -> v4: max positions 3 -> 4 (user, reason: adding HYPE)
11:40 UTC  incident INC-20260817-01 send timeout on HG-20260817-02; not on exchange, waited out expiresAfter 11:41, re-checked clean, re-approved and sent 11:47
18:00 UTC  note                    maker entries at 10 bps depth filled within 2h on both attempts this week

Rules: append only; corrections are new lines with correction:; every line has a UTC time and an id where one exists; no opinions in the journal (those go in reviews).

2. Trade review

Trigger: a proposal reaches closed, or the user asks. Inputs, always listed with timestamps:

  • the proposal file (ticket, PASS, approval, execution, reconciliation)
  • userFills / userFillsByTime for the window: price, size, fee, side, crossed
  • historicalOrders and orderStatus by cloid: what rested when, what cancelled
  • userFunding for the holding window
  • optionally the Market Analyst's depth read at send time for expected slippage

Compute:

MeasureHow
Entry slippage(fill avg - ticket price) / ticket price in bps, signed against the trade
Exit slippagesame for the exit versus its ticket or trigger price
Feessum of fill fees in USD and as bps of notional; note maker vs taker
Fundingsum of funding payments over the window, USD
Net resultrealised PnL after fees and funding, in USD and in R (R from the ticket)
Protectionwas a reduce-only stop resting on the exchange for the entire life of the position? gaps in minutes
Lifecycleeach stage present, in order, with timestamps; approval by id; single send; reconciled
Holding timefill to flat

Grade:

  • Process: clean / minor break / major break, with the specific stage named. A major break is any send without PASS or approval by id, any missing protection, any resend on unknown result, or any limit breached.
  • Outcome: result in R and USD, stated without adjectives.

Then one thing: a leak, a control that worked, or a break, chosen because it is repeatable. Not a list.

Write the review under ## review in the proposal file and send the block (format in agents/trade-reviewer.md) by DM to the Desk Lead and the user. Set status: closed.

3. Weekly desk review

From the journal, proposals and the exchange record for the week:

  • proposals opened / rejected / voided / executed / closed
  • trades closed: count, hit rate, average win and loss in R, expectancy in R (mean of results), largest loss, largest drawdown in equity terms from portfolio or start/end equity
  • costs: fees and funding in USD and as a share of gross PnL
  • process: number of breaks by type; incidents and their status
  • limits: any changes and why
  • one pattern worth the user's attention, stated as a fact pattern

No recommendations about what to trade. If the user asks, hand strategy questions to the Strategist and sizing questions to the Risk Manager.

4. Incident review

For each INC-YYYYMMDD-NN: timeline (journal + exchange record), what the desk did, what the controls did, exposure during the incident, root cause where knowable, one corrective action with an owner and a date. Blameless in tone, exact in fact. Written to /workspace/trading-desk/journal/incidents/<id>.md and linked from the daily journal.

Pitfalls

  • Reviewing from chat memory rather than fills. Chat says what people meant; fills say what happened.
  • Letting the outcome colour the process grade. Grade process first, then look at the outcome.
  • Counterfactuals ("if we had held..."). Not evidence.
  • Ten findings per review. Nobody acts on ten.
  • Editing old journal lines. Append a correction.

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.