Skip to content

CHANGELOG.md

Latest commit

 

History

History
89 lines (73 loc) · 6.3 KB

CHANGELOG.md

File metadata and controls

89 lines (73 loc) · 6.3 KB
Raw

Changelog

All notable changes to Aegis will be documented in this file.

[1.4.0] - 2026-03-31

Added

  • Consumer-Driven Integration Testing (Layer 4 upgrade) — integration tests are now driven by the frontend consumer, not just the backend provider. Cross-references frontend API calls with backend routes to catch missing endpoints before they reach production.
  • Route Manifest (contracts/route-manifest.yaml) — new Layer 2 contract artifact declaring every API route the frontend consumes. Auto-generated by scanning frontend code or manually maintained.
  • scripts/verify-route-coverage.sh — extracts frontend API calls (TypeScript/JavaScript, Go, Python), cross-references with backend routes, reports coverage gaps. Supports: monorepo scanning, manifest-based verification, and cross-workspace mode.
  • Consumer Route Manifest Exchange Protocol — cross-workspace protocol where frontend agent exports consumer-routes.yaml to the contract repo, backend CI validates coverage against it.
  • Three-tier verification: Full (cross-reference), Degraded (provider-only with warning), Error (missing backend).
  • Degradation strategy (铁律) — when frontend API surface is unavailable, falls back to provider-driven tests with ⚠️ WARNING (exit 0), never blocks CI.
  • Frontend API Surface section in Design Brief template — maps every frontend API call to its backend handler.
  • templates/route-manifest-starter.yaml — starter template for route manifests.

Changed

  • CI pipeline order updated: lint → type-check → unit → frontend-test → contract → integration → route-coverage → build → E2E
  • CLAUDE.md template: added consumer-driven route coverage testing requirement
  • CC skill synced with consumer-driven testing, route manifest, and verification script
  • Testing strategy reference: new Consumer-Driven Testing section with architecture-specific examples
  • Multi-agent protocol: new Consumer Route Manifest Exchange Protocol section

[1.3.1] - 2026-03-30

Improved

  • Testing strategy reference — complete rewrite with concrete code examples:
    • Frontend: Vitest + RTL + MSW patterns (API client test, hook test, component test, MSW handler setup)
    • Backend: Go and TypeScript/Node integration test examples with real HTTP server + real DB
    • Language-specific tool recommendations (Go, TS, Python, Rust, Java)
    • Coverage matrix template for backend endpoints
  • Design Brief template — expanded Testing Strategy section with structured tables:
    • Frontend test coverage checklist (API clients, hooks, components, MSW)
    • Backend endpoint coverage matrix (200/400/404/401/mutation)
    • DB setup requirements
  • Testing pyramid updated with all 5 layers (Unit → Frontend → Contract → Integration → E2E)
  • CI pipeline order standardized across all docs: lint → type-check → unit → frontend-test → contract → integration → build → E2E
  • Added principle #7: "Mock only what you don't own"

[1.3.0] - 2026-03-30

Added

  • Frontend Testing Standard — mandatory Vitest + React Testing Library + MSW stack for frontend projects. Coverage requirements for API clients, data hooks, key components. MSW handlers must use contract-typed response data. CI gate: pnpm test must pass.
  • Backend HTTP E2E Testing Standard — every API endpoint must have HTTP-level integration tests against real server + real database. Per-endpoint coverage: 200/400/404/401 + mutation verification. No mocking internal dependencies.
  • Test Strategy as Design Gate — full-stack features require complete testing strategy (frontend + backend + E2E) in the Design Brief before code begins. Design Review blocks approval without it.

Changed

  • Updated testing pyramid to include frontend test layer explicitly
  • CI pipeline order updated: lint → type-check → unit → frontend-test → contract → integration → build → E2E
  • CC skill Phase 4 updated with same standards

[1.2.0] - 2026-03-28

Added

  • Workspace Architecture Detection (CC skill) — Phase 0 auto-detection of project architecture before entering Aegis workflow. Detects monorepo/split/cross-agent and prompts human when ambiguous.
  • Cross-Workspace Contract Protocol (CC skill) — dedicated contract repository pattern, three integration methods (submodule/package/copy-sync), cross-workspace dispatch template, contract test isolation strategy.
  • Updated multi-agent protocol reference with full cross-workspace section.

Fixed

  • CC skill (cc-skill/) now includes all v1.1.0 features that were previously only in the OpenClaw SKILL.md.

[1.1.0] - 2026-03-28

Added

  • Workspace Architecture Detection — new phase before the Aegis workflow. Auto-detects project architecture (Monorepo / Split Workspace / Cross-Agent Cross-Workspace) by scanning directory structure and framework indicators. When ambiguous, prompts the human to confirm.
  • Cross-Workspace Contract Protocol — dedicated section in references/multi-agent-protocol.md for projects where frontend and backend live in separate workspaces managed by different agents. Covers:
    • Dedicated contract repository pattern
    • Three integration methods: Git submodule, package registry, copy-sync by lead
    • Dispatch prompt adjustments for cross-workspace agents
    • Contract sync workflow
    • Test isolation strategy (each side runs contract tests independently)
  • Architecture Mode Effects — different contract strategies and CLAUDE.md adjustments per detected mode
  • This CHANGELOG

[1.0.0] - 2026-03-27

Added

  • Five-layer protection system: Design → Contract → Implementation → Verification → PM
  • Layer 0: Automated guardrails (pre-commit hooks + CI pipeline, language-adaptive)
  • Layer 1: Design Brief template and workflow
  • Layer 2: Contract-first development (OpenAPI spec, shared types, error codes)
  • Layer 3: Implementation constraints via enhanced CLAUDE.md
  • Layer 4: Verification pyramid (unit → contract → integration → E2E)
  • Layer 5: Project management integration (gap tracking, sprint subtasks)
  • Multi-agent coordination protocol
  • CC Skill (Claude Code side) for persistent Aegis awareness
  • Scripts: init-project, setup-guardrails, detect-stack, validate-contract, generate-types
  • Self-contained CC skill package (cc-skill/) for pure Claude Code users
  • Lite Mode vs Full Mode support