Some checks failed
CI / Frontend Lint (pull_request) Failing after 8s
CI / Frontend Type Check (pull_request) Failing after 6s
CI / Frontend Build (pull_request) Failing after 5s
CI / Frontend E2E Tests (pull_request) Failing after 6s
CI / Orchestrator Build (pull_request) Failing after 6s
CI / Contracts Compile (pull_request) Failing after 6s
CI / Contracts Test (pull_request) Failing after 5s
Code Quality / SonarQube Analysis (pull_request) Failing after 18s
Code Quality / Code Quality Checks (pull_request) Failing after 5s
Security Scan / Dependency Vulnerability Scan (pull_request) Failing after 3s
Security Scan / OWASP ZAP Scan (pull_request) Failing after 4s
Closes gap-analysis v2 §4 partial (canonical "Execution Reference Set") and §10.6 SWIFT message ID persistence. - Migration 006 adds swift_message_id + swift_message_type columns to executions, with a partial index on swift_message_id for acknowledgment ingest (camt.025/054 -> original MT760/MT202 lookup). - db/executions.ts: recordExecution() UPSERT helper, getExecution(), findBySwiftMessageId() — the three queries the bank-instruction phase and SWIFT gateway need. - services/bank.ts.commitBankInstruction now emits a SWIFT reference alongside the ISO-20022 envelope: MT760 for plans carrying an issueInstrument step (real generateMt760 output, messageReference field), MT202 for payment-only plans (synthetic ref). - services/execution.ts persists the reference set at bank_instruction complete-time via recordExecution (best-effort; logs on failure, does not abort the leg). - 5 unit tests covering MT760 vs MT202 branching, reference uniqueness across calls, and SQL shape of the UPSERT + SELECT.