shredbx logo
shredbx shredbx shredbx shredbx Personal
  • Home
  • Lab
  • Portfolio
  • Experience
  • Services
  • Profile
  • Contact
AClaude
  • Home
  • Lab
  • Portfolio
  • Experience
  • Services
  • Profile
  • Contact
Andrei Solovev
Knowledge
Search knowledge... ⌘K
Knowledge · Rules · Process

Definition Of Done

Completion criteria — evidence-based done: UI visible, tests passing, fixtures present, no silent claims

Andrei Solovev

Tags

rule

Overview

Purpose

Completion criteria — evidence-based done: UI visible, tests passing, fixtures present, no silent claims

Rules

DOD-001: Work is NOT done unless visible in the application UI. Every artifact must have UI representation.

Invisible work is untestable by users and unverifiable by reviewers. UI visibility is the minimum evidence threshold.

Verification: UI check: navigate to the feature in the browser. If it's not visible, it's not done.

DOD-002: Source code deliverables REQUIRE fixtures: success/, failure/, edge-cases/ in fixtures/ folder.

Fixtures are executable documentation. They prove the code handles all scenarios, not just the happy path.

Verification: Directory check: fixtures/ exists with appropriate subdirectories for the deliverable.

DOD-003: Fixtures must be visible and browsable in the application UI for development and debugging.

Fixtures buried in test directories are invisible to developers during debugging. UI browsability makes them discoverable.

Verification: UI check: fixtures are navigable and renderable in the application.

DOD-004: Every deliverable requires tests: unit tests for logic, e2e tests for integration, Playwright for UI.

Tests are the acceptance criteria in executable form. No tests means no verifiable acceptance criteria.

Verification: Test check: unit, e2e, and UI tests exist and pass for the deliverable.

DOD-005: LLM self-validation via Playwright: verify work appears correctly in the UI before claiming done.

LLM agents cannot see the UI without Playwright. Without self-validation, claims are based on code expectation, not observed reality.

Verification: Playwright session: screenshot evidence of correct rendering stored in task artifacts.

DOD-006: Schema deliverables require Go validation scripts. No schema without validator.

Schemas without validators are just suggestions. Validators make schema compliance enforceable by CI.

Verification: Validator exists and passes for every schema deliverable.

DOD-007: Non-source-code deliverables use governance fields or comments as 'fixtures' for validation.

Not all deliverables have test suites. Governance fields (required, enforcement) serve as machine-checkable validation.

Verification: Governance fields populated for non-code deliverables.

DOD-008: All models controlled by Go modules. The application reads from modules, not raw YAML files.

Go modules provide type safety, validation, and versioning. Raw YAML parsing bypasses all compile-time guarantees.

Verification: Code check: YAML data accessed through Go module types, not raw map[string]interface{}.

DOD-009: Never claim work done without evidence. Evidence = UI visibility + tests passing + fixtures present.

Claims without evidence are assumptions. The three-part evidence requirement prevents premature completion claims.

Verification: Completion report includes: screenshot/URL, test output, fixture listing.

shredbx logo shredbx shredbx shredbx shredbx Andrei Solovev

Solution Architect & Lead Software Engineer

ExperiencePortfolioResearch & ExperimentsEducationCertificationSkills
GitHub ↗LinkedIn ↗Email ↗
AVAILABLE FOR NEW PROJECTS
// MY LATEST BEATS
Hobby & Interests

Lab

  • The Lab
  • Framework
  • Components
  • Packages
  • Games
  • Process (SDLC)
  • Knowledge
  • Blog

Andrei

  • Portfolio
  • Experience
  • Services
  • Profile
  • Contact
  • Lifestyle

Team

  • Team
  • Andrei
  • Claude

Legal

  • Privacy
  • Terms
  • Cookies
© 2026 shredbx.com. All rights reserved. — Andrei Solovev |