Skip to main content
reopt Handbook
reopt Handbook
Harness Engineering

Problem and Structure

Foundations of Harness EngineeringRepo-Readable SystemsThe Five Elements of a HarnessEngineering Mechanics

Evaluation and Interpretation

Evaluation Loop DesignExternal Case Comparison

Case Studies

Case: OpenAICase: AnthropicCase: TossCase: gstackCase: revfactory/harness

Domain Playbooks

Domain PlaybooksScenario: Frontend TeamScenario: Platform TeamScenario: Payments and Settlement TeamScenario: AI Product TeamWhy You Eventually Need Your Own Harness

Rollout and Operations

Team Harness Rollout StrategyTeam Harness Design ChecklistOperations: Entropy and Garbage Collection

Appendix

Verification ReportUpdates
Handbook›Harness Engineering›Scenario: Payments and Settlement Team
한국어English

Scenario: Payments and Settlement Team

Design a harness where approval, reconciliation, audit trail, and rollback reduce financial correctness risk.

Key takeaways

  • A payments harness treats "looks right" as a financial risk and makes approval policy, reconciliation, audit trail, and rollback the load-bearing elements.
  • Risk classification routes changes: operations docs auto-merge, fixture or settlement UI changes need review, and pricing, refund, or tax-field logic hits a human gate.
  • Reconciliation checks compare input fixture, expected ledger output, and actual persisted result, failing on minor-unit, tax-rounding, or refund-state mismatches.
  • The loop classifies risk before implementation, runs fixture and reconciliation checks, then requires human approval before deploy and a post-deploy audit.
  • First 30 days: attach risk-classification.yaml to pricing and settlement work, require a reconciliation-report.md before merge, and wire human gates into workflow, not just docs.

Payments and settlement teams are among the riskiest places to operate without a harness. "Looks right" can become direct financial loss.

Problem Structure

  • Small money-logic errors can cause loss.
  • Failures may appear only at settlement time.
  • Audit logs and approval evidence are needed later.
  • Risky changes must be identified before deployment.

Load-Bearing Elements

ElementWhy it matters
Approval policySeparates risky changes from automation
Reconciliation checkConfirms correctness, not just execution
Audit trailReconstructs who approved what and why
Rollback planLimits damage after a mistake

Recommended Loop

Artifact Structure

pricing-rules.md
ledger-invariants.md
approval-policy.md
risk-classification.yaml
reconciliation-report.md
audit-log.md
rollback-plan.md

Approval Policy Example

payments_change:
  auto:
    - "operations docs edits"
  review_required:
    - "test or fixture changes"
    - "settlement UI changes"
  human_gate:
    - "pricing logic changes"
    - "settlement rule changes"
    - "refund or cancellation logic changes"
    - "accounting or tax fields"

Reconciliation Example

reconciliation:
  compare:
    - "input fixture"
    - "expected ledger output"
    - "actual persisted result"
  fail_if:
    - "minor unit mismatch"
    - "tax rounding mismatch"
    - "refund state transition mismatch"

Why This Is Engineering

A payments harness is financial correctness engineering. It uses fixtures and ledger invariants, separates human gates, preserves audit evidence, and includes rollback as part of the design.

First 30 Days

  1. Attach risk-classification.yaml to pricing, refund, and settlement work.
  2. Require reconciliation-report.md before merge.
  3. Connect human gates to workflow, not only docs.

Read Next

  • /en/books/harness-engineering/case-toss
  • /en/books/harness-engineering/case-anthropic
  • /en/books/harness-engineering/operations

Related docs

Domain Playbooks

Translate harness principles into frontend, platform, payments, and AI product teams.

Why You Eventually Need Your Own Harness

Explain what to copy from external harnesses and what must be redesigned for your domain and team.

Scenario: Platform Team

Design a harness around invariants, impact analysis, shared modules, release gates, and migration discipline.

Scenario: AI Product Team

Design a harness where eval sets, safety policy, online telemetry, and model rollout control nondeterministic behavior.

On this page

Problem StructureLoad-Bearing ElementsRecommended LoopArtifact StructureApproval Policy ExampleReconciliation ExampleWhy This Is EngineeringFirst 30 DaysRead Next