v0.48.0

CLI Monolith Split & Community Skills

CLI Architecture Refactoring

• Split src/api/cli.ts from 6946 lines to 1660 lines (76% reduction)
• Extracted 16 command modules to src/cli/:
  • engineBootstrap.ts — engine singleton, shared helpers
  • sessionCommands.ts — session start/end
  • gateInlineCommands.ts — pre-tool/post-tool/before-stop hooks
  • metaGovernanceCommands.ts — G9-G15 meta governance gates
  • artifactCrudCommands.ts — create/list/show/suggest/createPRD
  • transitionCommands.ts — transition/verifyTask/role
  • diagnoseHuntCommands.ts — diagnose/hunt proactive scans
  • dependencyTddCommands.ts — dependency audit/TDD/stats/metrics
  • contextCommands.ts — context build/status/inject/glossary/budget
  • codegraphCommands.ts — code intelligence status/init/query/impact
  • evalCommands.ts — eval init/run/report/compare/replay
  • evolveDoctorCommands.ts — evolution cycle and doctor diagnostics
  • workflowEvidenceCommands.ts — workflow presets and evidence
  • initConfigCommands.ts — init/bootstrap/setup/config
  • upgradeAssetsCommands.ts — upgrade/assets/standards/artifact
  • runtimeSkillCommands.ts — runtime/memory/outOfScope/skill/token
  • toolAgentCommands.ts — tool/agent/team orchestration

Community Skills Integration

• Integrated 4 community skills into ExternalSkills registry:
  • qiushi-skill — materialist dialectics methodology (3.2k stars)
  • pua — high-agency persistent problem-solving (17.9k stars)
  • nuwa-skill — cognitive framework distillation (22.6k stars)
  • agency-agents-zh — 215 expert agent roles, 50 China-market (13.8k stars)

Phase 1 — Credibility Foundation

• ESLint: enable real TypeScript lint rules (1748 warnings, 0 errors)
• G0-verify.sh: structural integrity check (package.json, tsconfig, cli.ts)
• Version drift prevention test added

Community & Documentation

• Added QQ group (628043364) and email (2496155694@qq.com) to README and COMMUNITY.md
• Added Feishu group, WeChat ID, and WeChat public account QR codes
• Updated English README badges and version references

---

Full Changelog: v0.47.0...v0.48.0

v0.47.0

Improvement Roadmap: 8 Items Implemented

P0-1: Fast-lane Mode

• Added PREFLIGHT_FAST_LANE_GATES constant (G0, G3, G4, G5) for S-level task acceleration
• Extended preflightGateStages() with 'fast-lane' profile support
• Shell runner (scripts/gates/all.sh) now accepts --fast-lane flag
• Per-gate timing added to gate runner output

P0-2: Upgrade Automation

• createUpgradeRecommendReport(): risk scoring (low=1, medium=3, high=5, blocker=+5), recommendation types (safe-to-apply / review-first / blocked)
• createGitBackup(): auto-stash, backup branch creation, restore
• upgrade recommend CLI command with --auto-apply flag
• upgrade apply --auto-backup for git-based rollback safety
• Full troubleshooting guide: docs/guides/UPGRADE_AUTOMATION.md

P1-5: Learning Path & Onboard Wizard

• Onboard wizard: 4 questions (experience, project-stage, team-size, priority) → profile recommendation with confidence scoring
• Profiles: minimal, standard, advanced, china-local
• scale onboard CLI command (interactive + --json mode)
• Learning path: 5 levels (Explorer → User → Configurator → Governor → Contributor)
• Guide: docs/guides/LEARNING_PATH.md

P1-7: Performance Baseline

• Measurement script: scripts/performance/measure-gates.sh (multi-run CSV)
• Baseline doc: docs/PERFORMANCE_BASELINE.md (G3=281ms, G4=1755ms, G5=467550ms)
• CI workflow: .github/workflows/performance-baseline.yml (weekly + manual)

P2-8: Multi-Agent Coordination (G13) Enforcement

• G13 now reads .scale/coordinator/state.json for real coordination evidence
• Checks activeSessions, overlaps, open conflicts → blocks on open conflicts
• Shell script: scripts/gates/G13-verify.sh (agent config, coordinator state, events)

P3-10: Token Budget (G21) Enforcement

• G21 ContextBudgetGate now adds blockers[] when tokens exceed budget
• Gate passed field is now dynamic (based on blockers length)
• Shell script: scripts/gates/G21-verify.sh exits non-zero on over-budget

P3-11: Session Health (G22) Fine-grained Signals

• G22 SessionHealthGate added Check 4 (.scale directory size via du -sk, warn >100MB)
• G22 SessionHealthGate added Check 5 (disk space via df -k, warn <1GB)
• Shell script: scripts/gates/G22-verify.sh with matching checks

P3-12: Documentation Link Hygiene (G17) Enforcement

• G17 DocumentationHygieneGate broken links now added to blockers[] (was advisory)
• Gate marked blocking: true in catalog
• Shell script: scripts/gates/G17-verify.sh exits non-zero on broken links

---

Full Changelog: v0.46.0...v0.47.0

v0.46.0

Tracker Adapters, SPA Components, VSCode Extension & Eval Benchmarks

Tracker Adapters

• Added LinearTrackerAdapter — full Linear GraphQL API integration (fetchCandidates, updateState, addComment, getIssue)
• Added JiraTrackerAdapter — full Jira REST API v3 integration with transition-based state management
• Wired both adapters into OrchestratorDaemon.createTracker() (env vars: LINEAR_API_KEY, JIRA_API_TOKEN, JIRA_BASE_URL)
• Added 18 unit tests for Mock/Linear/Jira tracker adapters

SPA Components

• Added 6 reusable dashboard components: MetricCard, Panel, StatusBadge, DataTable, EventStream, LoadingState
• Added component CSS (badges, data tables, loading spinners, empty states, event stream)
• Integrated component scripts into index.html

VSCode Extension

• Added vscode-extension/ — full VS Code extension scaffold
• 7 commands: status, verify, context, dashboard, eval run, shield status, cortex evolve
• 3 Tree Views: Artifacts, Gates, Events
• Configuration: projectDir, dashboardPort, autoVerify

Eval Benchmarks

• Expanded workflow-baseline suite from 1 to 20 cases
• 10 bugfix, 5 feature, 3 refactor, 1 security, 1 smoke
• Covers: null safety, regex escape, JSON parse, event leak, path traversal, async timeout, concurrent write, empty array, UTF-8 BOM, API endpoint, CLI flag, webhook config, filter query, diff view, module extraction, complexity reduction, type consolidation, secret detection

Stub Code Elimination

• PhasePromptRegistry.loadCustom() — implemented .scale/prompts/*.md and ~/.claude/prompts/*.md custom prompt loading
• PhasePromptRegistry.exportPrompt() — implemented file write
• EvolutionEngine.HookGenerator — implemented pattern-specific shell checks (secret detection, dangerous commands, error handling, verification requirements)
• WorkflowOrchestrator.simulateExecution() — implemented OpenAI-compatible LLM API call with fallback

Test Coverage

• Added tests/guardrails/reviewEnforcer.test.ts — 15 tests for ReviewEnforcer (enforce, rollback, iteration limit)
• Added tests/guardrails/roles.test.ts — 11 tests for 6 role definitions
• Added tests/guardrails/detectorsExtended.test.ts — 7 tests for BlameShiftDetector, IdleToolDetector
• Added tests/hooks/hookDeployer.test.ts — 19 tests for HookDeployer (validate, deploy, rollback, events)
• Total: 70 new tests, 1599 tests passing

Full Changelog: v0.45.0...v0.46.0

v0.45.0

Optimization Plan Completion & Memory Provider Integration

5-Phase Optimization Plan

• Phase 1 - Release Hygiene: Version sync (v0.45.0), npm package slim (no images)
• Phase 2 - CLI Refactor + Gates: CLI split into src/cli/ modules, gates G16-G22 implemented
• Phase 3 - Documentation: CLI reference (docs/reference/cli.md), architecture diagrams (docs/architecture/README.md), README updated
• Phase 4 - E2E Verification: Shield status (inject/verify), Orchestrator status --json, Cortex verify
• Phase 5 - Version Migration: Migration guide (docs/migration/v0.38-to-v0.44.md), governance.lock.json compatibleVersions

3-Layer Memory Architecture

• Added MemoryNodeLayer type: L1-trace → L2-policy → L3-world-model → crystallized
• Added refine() method for knowledge extraction across layers
• SQLite schema migration from v1 to v2 with layer column
• Updated query() and list() to support layer filter parameter
• Updated dream() to include byLayer statistics

Memory Provider Integration

• Added MemOS (memtensor/MemOS) as memory provider
• Updated default provider order: gbrain → memos → agentmemory → scale-local
• Added contextSavings to MemoryProviderRecallReport for token savings estimation
• Added estimateTokens() helper for context size calculation

Code Intelligence

• Added code-review-graph (tirth8205/code-review-graph) as code intelligence provider
• Registered MCP tools as Skills: code-change-detection, code-review-context
• Added keyword matching in recommendSkillWorkflow() for review tasks

Hooks & Automation

• Added SessionStart hook with scale cortex inject --minimal and fallback
• Added CRG incremental update hook (PostToolUse) for automatic graph rebuild on file save

Documentation

• Added docs/reference/cli.md — complete CLI reference
• Added docs/architecture/README.md — 12 Mermaid diagrams (system overview, core engines, intelligence layer, workflow pipeline, data flow, storage architecture, integration points)
• Added docs/migration/v0.38-to-v0.44.md — migration guide covering breaking changes and new features
• Updated docs/EXTERNAL_REFERENCES.md with MemOS and code-review-graph entries
• Updated README.md badge versions to 0.45.0

Test & Quality

• Fixed memory provider order in 6 test files
• Added missing external references documentation
• All 167 test files / 1599 tests passing

Full Changelog: v0.43.0...v0.45.0

v0.44.0

Dashboard SPA & Code Topology

Dashboard SPA Frontend

• Added SPA frontend with 6 pages: overview, costs, documents, monitoring, topology, workflow.
• Added i18n support (English/Chinese).
• Replaced static dashboard server with Hono-based API backend.
• Added @hono/node-server and echarts dependencies.
• Added E2E verification scripts (verify-dashboard.mjs, verify-dashboard-browser.mjs).

Code Topology Visualization

• Added code topology visualization system (P1-P5).
• Added graphify nodes[]/links[] format support in topology dump.
• Fixed graphify manifest fallback in provider status check.

CI/CD

• Added GitHub Release creation to publish workflow.

Full Changelog: v0.43.0...v0.44.0

v0.43.0 — Governance Hardening & Migration Readiness

v0.43.0 — 2026-05-27

Governance Hardening & Migration Readiness

Enhanced Gate System (G16-G22)

• 7 new enhanced gates: G16 (lock sync), G17 (architecture), G18 (performance), G19 (error handling), G20 (documentation), G21 (test coverage), G22 (dependency hygiene)
• EnhancedGates.ts with GateActivation modes (default/profile/optional/policy)
• GateCatalog extended with GateFamily classification (core/meta/extension)

Migration & Deprecation

• docs/guides/MIGRATION.md — breaking changes from v0.26.0 through v0.42.0
• DeprecationWarning system in UpgradeManager with 4 deprecation entries
• scale upgrade plan now shows deprecation warnings for old versions

Governance Lock Sync

• Recalculated all 27 managed file SHA-256 hashes (was stale since v0.26.0)

Cortex Pipeline

• scale cortex verify command with 6 health checks

Bug Fixes

• DependencyAuditor: downgrade eval() severity for browser automation packages (playwright, puppeteer, jsdom, happy-dom)
• phaseCommands: fix gate status reporting — 'blocked' now correctly returned when gates block

Test Coverage

• Full suite: 164 files, 1572 tests, all passing

v0.39.0

0.39.0 - 2026-05-24

Prompt Optimization

• Added deterministic scale prompt optimize for structured coding prompts without model calls.
• scale define now optimizes requirements before ambiguity scoring and spec creation.

Gate Catalog and Task Scoring

• Added scale gates status --json for machine-readable gate catalog visibility.
• Added scale score task --changed --json for deterministic task scoring.
• Scoped engineering standards checks to Git-changed files inside Git worktrees.

Validation: npm run release:check passed, including typecheck, lint, full tests, setup smoke, build, audit, git diff --check, and npm pack --dry-run.