Skip to content

Latest commit

 

History

History
74 lines (53 loc) · 3.15 KB

File metadata and controls

74 lines (53 loc) · 3.15 KB

Finance Reconciliation Boundary Model

Purpose

This profile-testbed example describes the agentic boundaries for LedgerOps Intelligence.

It is not a repo-local production claim. The repository remains independent: it is not an AAIF project, not a Linux Foundation project, and not endorsed by either organization.

High-Risk Action Boundary

Financial account connection, transaction classification, reconciliation suggestions, exception handling, report export, and any action that changes financial records or payment state.

Actors

Actor Responsibility
Finance operator Reviews accounts, reconciliations, exceptions, and reports.
Account owner Grants or revokes account-data consent.
Finance agent Proposes classifications, matches, and explanations.
Banking or ledger adapter Reads account, transaction, and ledger data through approved scopes.
Maintainer Reviews consent, audit, and financial-control evidence.

Identity Boundary

  • Account consent must be distinct from application login.
  • The agent cannot own bank, ledger, or payment credentials.
  • Financial write actions require human operator identity and approval.
  • Exported reports must carry source, timestamp, and review status.

Tool And Action Boundary

Action Boundary
read_transactions Allowed only for consented accounts and approved date ranges.
classify_transaction Draft classification until reviewed or policy-autopassed.
propose_reconciliation Must preserve source records and confidence rationale.
export_finance_report Requires approval and redaction review.
initiate_payment Out of scope unless a separate high-risk approval system exists.

Human Approval Gates

  • Before connecting or reconnecting financial accounts.
  • Before accepting reconciliation suggestions above risk thresholds.
  • Before exporting reports outside the finance workspace.
  • Before changing ledger records, payment state, or account mappings.

Audit Event Model

Event Minimum Evidence
account.consent_granted account owner, scope, provider, expiration
transaction.classified transaction id, proposed class, confidence, reviewer
reconciliation.proposed source records, match reason, variance, risk class
reconciliation.approved approver, policy result, timestamp
finance_report.exported report id, redaction status, recipients, approval id

Local Proof Path

  • bun install to prove dependency resolution.
  • bun run lint to prove static project health.
  • bun run test for finance policy and reconciliation checks when available.
  • bun run build to prove the application compiles.
  • Use synthetic account and transaction data in public proof.

OSS And Self-Hosted Fallbacks

  • Use local CSV or OFX import before hosted banking aggregators for demos.
  • Use PostgreSQL or SQLite fixtures for transaction and audit data.
  • Keep banking, payment, and ledger providers behind adapters.
  • Use local model adapters for explanation drafts where feasible.

Contributor Follow-Up

Add a repo-local docs/AGENTIC_BOUNDARY_MODEL.md after review and link it from agentic-readiness.json.