Inputs

  • PROPOSED_FIRST_ACTION
  • VENUE_QUIRKSthe relevant venue's Quirks Registry entry
  • SUB_ACCOUNT_CONFIG

Output format

Dry-run report with issues flagged.

The prompt

ROLE: You are dry-running an agent's proposed first live action against
the venue's known quirks and the sub-account's configuration. Catch
issues that would manifest as production incidents.

PROPOSED ACTION. [PROPOSED_FIRST_ACTION]
VENUE QUIRKS. [VENUE_QUIRKS]
SUB-ACCOUNT CONFIG. [SUB_ACCOUNT_CONFIG]

REVIEW.
1. Does the proposed action format match the venue's API mode
   (registration vs submission)?
2. Are precision rules respected (price, size, leverage decimals)?
3. Does the proposed quote/order deadline allow for solver fill latency
   per the venue's known timing characteristics?
4. Does the proposed notional sit within solver depth at submission time?
5. Are there known venue quirks for the chosen instrument that the
   proposed action could trip?
6. Does the sub-account have sufficient margin?
7. Is the action consistent with the sub-account's tier (paper actions
   should not have real-capital expectations)?

OUTPUT.
{
  "verdict": "go" | "no-go" | "go-with-caveats",
  "issues": [
    { "category": "<one of the review categories>",
      "severity": "blocking" | "warn" | "info",
      "description": "<specific>",
      "remediation": "<specific>" }
  ],
  "summary": "<one paragraph>"
}

Usage notes

Run as the last step before the first action on any new sub-account. Most "go-with-caveats" verdicts surface issues the operator wants to know about but doesn't need to block on; "no-go" verdicts indicate a genuine integration mismatch that should be resolved before submission.

Open tier · MIT · v0.1 — the Pro library updates monthly, venue-tuned via the Quirks Registry. Templates, not advice.