viepilot 2.23.0 → 2.41.0

package/CHANGELOG.md

@@ -7,6 +7,201 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [2.41.0] - 2026-04-24
+
+### Added
+- **ENH-072** — vp-* skill invocation version update check
+  - `lib/viepilot-update.cjs`: `checkLatestVersion()` — fetches latest version from npm registry (`https://registry.npmjs.org/viepilot/latest`) with 3s timeout; caches result 24h in `~/.viepilot/update-cache.json`; silent on all failures (network error, timeout, parse error, unreadable cache)
+  - `bin/vp-tools.cjs`: `check-update` subcommand — `--silent` exits 1 + prints latest when update available (exits 0 when up-to-date); `--json` prints `{ installed, latest, has_update }`; `--force` bypasses 24h cache
+  - All 18 `skills/vp-*/SKILL.md` files: `<version_check>` block inserted after `<greeting>` — runs `vp-tools check-update --silent` after the greeting banner; shows notice banner if exit=1; suppressed by `--no-update-check` flag or `config.json update.check: false`; shown at most once per session via `update_check_done` guard
+
+## [2.40.0] - 2026-04-23
+
+### Added
+- **ENH-071** — vp-brainstorm Embedded Domain Mode (10 gaps addressed)
+  - Auto-activates when ≥2 embedded trigger keywords detected (MCU families + firmware concepts); `--domain embedded` flag for manual activation
+  - `workflows/brainstorm.md`: `detect_embedded_domain` step (Step 0B) with MCU/concept keyword lists, activation banner, and domain flag
+  - Embedded topic probes injected when domain active: MCU/Toolchain (Gap 2), RTOS/Scheduling (Gap 3), Power Budget (Gap 7), Safety/Compliance (Gap 10), Firmware Phase Template (Gap 9)
+  - 6 new Architect workspace pages: `hw-topology.html`, `pin-map.html`, `memory-layout.html`, `protocol-matrix.html`, `rtos-scheduler.html`, `power-budget.html` (Gaps 1, 4, 5, 8); linked under Embedded nav section in `index.html`
+  - UI Direction false-positive suppression (Gap 6): hardware display keywords (LCD/OLED/TFT) routed to `hw_topology_buffer` instead of `ui_idea_buffer`; `🎨` banner suppressed when all signals have hardware context
+  - `workflows/crystallize.md`: Step 1D item 13 — exports 8 embedded YAML sections (`hw_topology`, `pin_map`, `memory_layout`, `protocols`, `rtos_config`, `embedded_toolchain`, `power_budget`, `safety_config`) to ARCHITECTURE.md hardware sections; `## Embedded Domain` written to PROJECT-CONTEXT.md; hardware sections READ-ONLY for vp-auto
+  - `workflows/crystallize.md`: Step 7 embedded path — skips UI Coverage Gate; applies Hardware Coverage Check (driver-task completeness warning, non-blocking) + Board Bring-Up Phase 1 auto-add
+  - `skills/vp-brainstorm/SKILL.md`: Embedded Domain Mode section with all 6 pages table, topic probes summary, suppression behavior, crystallize exports summary
+
+## [2.39.1] - 2026-04-23
+
+### Fixed
+- **BUG-021** — Antigravity adapter install path updated for Google Gemini ecosystem rebrand
+  - `lib/adapters/antigravity.cjs`: `skillsDir` → `~/.gemini/antigravity/skills/`
+  - `lib/adapters/antigravity.cjs`: `viepilotDir` → `~/.gemini/antigravity/viepilot/`
+  - `lib/adapters/antigravity.cjs`: `executionContextBase` → `.gemini/antigravity/viepilot`
+  - `lib/adapters/antigravity.cjs`: `isAvailable` checks `.gemini/antigravity/` with `.antigravity/` fallback for pre-update installs
+  - `bin/viepilot.cjs`: uninstall help text updated to `~/.gemini/antigravity/skills/vp-*`
+  - `docs/user/features/adapters.md`: Antigravity row paths updated
+  - `docs/user/features/skill-registry.md`: Antigravity row path updated
+
+## [2.39.0] - 2026-04-22
+
+### Added
+- **ENH-070** — vp-audit auto-log gaps as requests + direct vp-evolve routing
+  - `workflows/audit.md`: Auto-Log Gate fires after all tiers; each ⚠️/⛔ finding auto-creates `.viepilot/requests/{TYPE}-{N}.md`
+  - Duplicate detection: existing open requests with matching title/files are updated (re-detected) instead of creating duplicates
+  - TRACKER.md backlog updated automatically after audit
+  - Post-audit routing banner shows logged request IDs + recommends `/vp-evolve` as next action
+  - AUQ prompt: "Plan fix phase → /vp-evolve" as Recommended option
+  - `--no-autolog` flag disables auto-logging for report-only mode
+  - `skills/vp-audit/SKILL.md`: documents Auto-Log Behavior + `--no-autolog`
+
+## [2.38.0] - 2026-04-22
+
+### Added
+- **ENH-069** — crystallize UI→Task binding: 10-gap fix closing the chain where prototype pages remain stubs after all phases complete
+  - Gap 1: crystallize Step 1A emits `## UI Pages → Component Map` table binding each pages/*.html to a target component
+  - Gap 2: crystallize Step 7 cross-checks generated ROADMAP vs component map; auto-adds missing tasks
+  - Gap 3: `templates/phase/TASK.md` and `autonomous.md` populate `## UI Prototype Reference` for frontend component tasks
+  - Gap 4: `autonomous.md` phase completion UI Coverage Gate — warns on stub components before phase PASS
+  - Gap 5: crystallize Step 1D converts `arch_to_ui_sync` entries with `status: noted` into component map rows
+  - Gap 6: crystallize Step 1F makes coverage[] gaps blocking for scoped features (not just warnings)
+  - Gap 7: crystallize Step 1D requires explicit resolution of feature-map.html discrepancies before Step 7
+  - Gap 8: crystallize emits "design staleness" warning when arch decisions are not reflected in pages/*.html
+  - Gap 9: crystallize Step 1A processes `## UX walkthrough log` P0/P1 pain items → UX-fix tasks
+  - Gap 10: crystallize Step 1A surfaces `## Background extracted ideas` with resolution gate before Step 7
+
+## [2.37.0] - 2026-04-22
+
+### Fixed
+- **BUG-020** — `vp-auto` scaffold-first gate: when a project setup task is detected for a framework stack (Laravel, Next.js, NestJS, Rails, Django, Spring Boot, Nuxt, React, Electron), vp-auto now MUST run the canonical scaffold command (e.g. `composer create-project laravel/laravel`, `npx create-next-app@latest`) before creating any files. A never-handcraft block list prevents manual creation of framework-native files (`artisan`, `manage.py`, `next.config.*`, `nest-cli.json`, `pom.xml`, `config/application.rb`, etc.) without prior scaffold. Stack `SUMMARY.md` files support an optional `## Scaffold` section with `init_command:` and `marker_file:` fields for override.
+
+### Added
+- `workflows/autonomous.md`: `#### Scaffold-First Gate (BUG-020)` section — framework marker detection table, built-in scaffold command heuristic table (10 stacks), never-handcraft block list, waiver/bypass prompt
+- `docs/user/features/scaffold-first.md` — scaffold-first convention documentation and stack configuration guide
+- `## Scaffold` sections (with `init_command:` + `marker_file:`) added to 6 installed framework stack SUMMARY.md files: laravel, nextjs, nextjs-tailwind-shadcn-threejs, nestjs, spring-boot, spring-boot-3.4
+
+## [2.36.1] - 2026-04-22
+
+### Fixed
+- **BUG-019** — `vp-tools scan-skills` subcommand now implemented; previously threw "Unknown command: scan-skills" even though `lib/skill-registry.cjs` already contained `scanSkills()` from FEAT-020. Wired CLI handler, `help scan-skills` entry, and usage summary line in `bin/vp-tools.cjs`.
+
+## [2.36.0] - 2026-04-22
+
+### Added (ENH-068)
+- `workflows/brainstorm.md`: Topic 7 "Admin Entity Management" (10 sub-questions: which entities need admin CRUD, list view columns/filter/pagination, bulk ops, create/edit forms, delete semantics hard vs. soft, import/export CSV, audit trail per entity, multi-tenant scoping, read-only vs. editable); proactive 🗄️ heuristic (CRUD/entity/admin panel keywords); entity management coverage gate before /save (cross-references ERD presence); `entity-mgmt.html` Architect page; `notes.md ## entity_mgmt` YAML schema; Content Management → 8, User Data Management → 9, Phase assignment → 10
+- `workflows/crystallize.md`: Step 1D item 10 entity_mgmt export → `## Admin Entity Management` table + Import/Export sub-table in `PROJECT-CONTEXT.md`; `entity-mgmt.html` added to MANDATORY READ GATE as item 15; items 10→11 (feature-map), 11→12 (working notes)
+- `templates/project/PROJECT-CONTEXT.md`: `## Admin Entity Management` section + Import/Export placeholder sub-tables
+- `skills/vp-brainstorm/SKILL.md`, `skills/vp-crystallize/SKILL.md`: ENH-068 documented
+- `tests/unit/vp-enh031-language-standardization.test.js`: exemption for entity_mgmt keyword blockquote (`CRUD` pattern)
+
+## [2.35.0] - 2026-04-22
+
+### Added (ENH-067)
+- `workflows/brainstorm.md`: Session template adds `workflow_version` + `upgrade_supplement_version` fields to `## Session Info`; **Step 3B: Upgrade Gap Detection** — on `--continue`, compares session `workflow_version` vs. current Topics Template using version threshold table (ENH-063 v2.32, ENH-065 v2.33, ENH-066 v2.34) with false-positive cross-check; shows 🔄 upgrade banner (AUQ: discuss now / remind at /save / skip); runs inline Q&A for missing topics; appends `## Upgrade supplement (vX → vY)` section; stamps `upgrade_supplement_version` for idempotency; suggests `/vp-crystallize --upgrade`
+- `workflows/crystallize.md`: **Step 0-B: Upgrade Re-scan Mode** (`--upgrade` flag + auto-detect from `crystallize_version` stamp); delta computation from version threshold table; AUQ menu — **Patch** (non-destructive: reads architect YAML → appends missing sections, re-stamps version) / **Full re-generate** (backup `.viepilot/backup-pre-regen-{ts}/` → pre-fill Step 0 → overwrite artifacts) / **Skip**; brainstorm supplement integration (folds `## Upgrade supplement` content automatically); soft warn when architect `notes.md` YAML missing; Step 5 writes `<!-- crystallize_version: {semver} -->` as first line of `PROJECT-CONTEXT.md`; `HANDOFF.json` records `crystallize_version` + `crystallized_at`
+- `skills/vp-brainstorm/SKILL.md`: Workflow version stamps + Upgrade gap detection bullets (ENH-067)
+- `skills/vp-crystallize/SKILL.md`: Crystallize version stamps + Upgrade re-scan mode bullets (ENH-067)
+
+## [2.34.0] - 2026-04-21
+
+### Added (ENH-066)
+- `workflows/brainstorm.md`: Topic 8 "User Data Management" in brainstorm Topics Template (10 sub-questions: profile management, notification preferences, privacy/GDPR controls, activity history, connected OAuth accounts, session/device management, 2FA, consent management, data retention); proactive 👤 heuristic (≥1 user-data keyword at session start, ≥2 during session); user data coverage gate before /save for B2C/SaaS/GDPR projects; `user-data.html` page in Architect workspace; `notes.md ## user_data` YAML schema; Phase assignment renumbered to Topic 9
+- `workflows/crystallize.md`: Step 1D extended — exports `notes.md ## user_data` → `## User Data Management` capabilities table in `PROJECT-CONTEXT.md`; `user-data.html` added to MANDATORY READ GATE list as item 14
+- `templates/project/PROJECT-CONTEXT.md`: `## User Data Management` section with placeholder capability table (8 rows: profile editing, notification prefs, privacy settings, data export, right to erasure, connected accounts, session management, 2FA)
+- `skills/vp-brainstorm/SKILL.md`, `skills/vp-crystallize/SKILL.md`: ENH-066 documented
+
+## [2.33.0] - 2026-04-21
+
+### Added (ENH-065)
+- `workflows/brainstorm.md`: Topic 7 "Content Management" in brainstorm Topics Template (10 sub-questions: content types, lifecycle, WYSIWYG/markdown, media, taxonomy, localization, search, versioning, SEO); proactive 🗂️ heuristic (≥1 content keyword at session start, ≥2 during session); content coverage gate before /save; `content.html` page in Architect workspace + Page Boundary Rules; `notes.md ## content` YAML schema; Phase assignment renumbered to Topic 8
+- `workflows/crystallize.md`: Step 1D extended — exports `notes.md ## content` → `## Content Management` table (content types, media/storage, localization) in `PROJECT-CONTEXT.md`; `content.html` added to MANDATORY READ GATE list
+- `templates/project/PROJECT-CONTEXT.md`: `## Content Management` section with Content Type, Media & Storage, and Localization placeholder tables
+- `skills/vp-brainstorm/SKILL.md`, `skills/vp-crystallize/SKILL.md`: ENH-065 documented
+
+## [2.32.0] - 2026-04-21
+
+### Fixed (BUG-018)
+- `workflows/brainstorm.md`: added unified workspace mode-selection AUQ prompt (Step 2B) after scope lock — user now chooses Both / Architect only / UI Direction only / Neither; Architect auto-activate heuristic deferred until after this selection; suppressed when user selects "Neither" or "UI Direction only"
+- `skills/vp-brainstorm/SKILL.md`: objective updated with BUG-018 mode selection reference
+
+### Added (ENH-063)
+- `workflows/brainstorm.md`: Topic 6 "Admin & Governance" in brainstorm Topics Template; proactive 🔐 heuristic (≥1 admin keyword at session start, ≥2 during session); admin coverage gate before /save for multi-user/SaaS/compliance projects; `admin.html` page in Architect workspace; `notes.md ## admin` YAML schema
+- `workflows/crystallize.md`: Step 1D extended — exports `notes.md ## admin` → `## Admin & Governance` table (capabilities + personas) in `PROJECT-CONTEXT.md`
+- `templates/project/PROJECT-CONTEXT.md`: `## Admin & Governance` section with placeholder table
+- `skills/vp-brainstorm/SKILL.md`, `skills/vp-crystallize/SKILL.md`: ENH-063 documented
+
+### Added (ENH-064)
+- `workflows/brainstorm.md`: cross-workspace HUB links — Architect `index.html` ↔ UI Direction `index.html` when both workspaces active in same session; triggered by arch_to_ui_sync, BUG-018 "Both" selection, or `/sync-links`
+- `workflows/crystallize.md`: Step 1D MANDATORY READ GATE — if architect workspace exists, reads ALL 12 pages front-to-back before extraction; `architect_read_complete: true` required; `notes.md` missing → STOP
+- `workflows/crystallize.md`: Step 1A strengthened — if ui-direction exists, reads ALL pages/*.html + ALL notes.md sections; `ui_direction_read_complete: true` required; pages inventory mismatch → STOP
+- `workflows/crystallize.md`: Step 1F Cross-Reference Gate — validates `## Coverage` matrix when both workspaces present; warns on Phase 1 features with no architect OR UI coverage
+- `skills/vp-crystallize/SKILL.md`: mandatory workspace read gates documented
+
+## [2.31.0] - 2026-04-21
+
+### Added (ENH-062)
+- `skills/vp-skills/SKILL.md`: new `/vp-skills` slash command — agent-native global skill registry management from any project; commands: scan, list, install, uninstall, update, info; uses installed vp-tools path (`~/.claude/viepilot/bin/vp-tools.cjs`)
+- `bin/vp-tools.cjs`: new `get-registry [--id <id>]` subcommand — outputs registry JSON to stdout; enables workflow shell integration
+
+### Fixed (BUG-016)
+- `workflows/autonomous.md`: replaced non-executable `Call loadRegistry()` with `node ~/.claude/viepilot/bin/vp-tools.cjs get-registry --id {id}` shell command
+- `workflows/brainstorm.md`: replaced `Call loadRegistry()` with `node ~/.claude/viepilot/bin/vp-tools.cjs get-registry` shell command + cat fallback
+
+### Fixed (BUG-017)
+- `skills/vp-evolve/SKILL.md`: added explicit AUQ call in Step 5 `<process>` body — workflow continuation prompt (Execute /vp-auto / Create request / Done)
+- `skills/vp-request/SKILL.md`: added explicit AUQ call in Step 6 `<process>` body — workflow continuation prompt (Plan /vp-evolve / Create request / Done)
+
+## [2.30.0] - 2026-04-20
+
+### Changed (FEAT-020 Phase 5 — complete)
+- `workflows/autonomous.md`: Added **Skill Context Load** step at init — reads `PROJECT-CONTEXT.md ## Skills`, builds `SKILL_CONTEXT_MAP`; absent section → silent no-op
+- `workflows/autonomous.md`: Added **Skill Context Injection** per-task rule — required skills whose phases match are prepended silently to execution context; records `skills_applied` in task output; never prompts user
+- `skills/vp-auto/SKILL.md`: FEAT-020 silent skill execution docs
+- **FEAT-020 fully complete** (Phases 90–94): Skill Registry System — scanner + installer + brainstorm integration + crystallize gate + vp-auto execution
+
+## [2.29.0] - 2026-04-20
+
+### Changed (FEAT-020 Phase 4)
+- `workflows/crystallize.md`: Added **Step 1E — Skill Decision Gate** — reads `## skills_used` from brainstorm notes.md; AUQ confirm required/optional/exclude; writes `## Skills` to PROJECT-CONTEXT.md; silently skipped when no skills_used present
+- `templates/project/PROJECT-CONTEXT.md`: Added `## Skills` template section
+- `skills/vp-crystallize/SKILL.md`: FEAT-020 skill gate docs — AUQ flow, lock semantics, cross-references
+
+## [2.28.0] - 2026-04-20
+
+### Changed (FEAT-020 Phase 3)
+- `workflows/brainstorm.md`: Added **Skill Registry Integration** step in UI Direction Mode — when active, loads `~/.viepilot/skill-registry.json`, matches skills by `capabilities` to UI signals, silently applies `best_practices` to HTML generation, records matched skills in `notes.md ## skills_used`
+- `skills/vp-brainstorm/SKILL.md`: FEAT-020 skill integration docs — silent apply, crystallize lock reference
+
+## [2.27.0] - 2026-04-20
+
+### Added (FEAT-020 Phase 2 — Skill Installer)
+- `lib/skill-installer.cjs`: new `installSkill(source, home?)`, `uninstallSkill(id, home?)`, `updateSkill(id, home?)` — multi-channel installer (npm / GitHub / local); writes `skill-meta.json` post-install; auto-runs `scanSkills()`
+- `bin/viepilot.cjs`: new `install-skill`, `uninstall-skill`, `update-skill` subcommands
+
+## [2.26.0] - 2026-04-20
+
+### Added (FEAT-020 Phase 1 — Skill Registry Foundation)
+- `lib/skill-registry.cjs`: new `scanSkills(home?)` — traverses all adapter `skillsDir` paths, parses SKILL.md (extended + legacy), deduplicates by id, writes `~/.viepilot/skill-registry.json`
+- `lib/skill-registry.cjs`: new `loadRegistry(home?)` — reads and returns registry JSON (null if absent)
+- `bin/viepilot.cjs`: new `scan-skills` subcommand — runs scanner, prints per-adapter summary + total unique skills
+- `bin/viepilot.cjs`: new `list-skills` subcommand — displays indexed skills table from registry
+- `docs/user/features/skill-registry.md`: extended SKILL.md format spec (`## Capabilities`, `## Tags`, `## Best Practices`), scanner behavior, registry JSON schema, commands table
+- Extended SKILL.md format: optional metadata sections — backward compatible (legacy SKILL.md without sections indexed with `[]` fields)
+
+## [2.25.0] - 2026-04-18
+
+### Changed (ENH-061)
+- `workflows/brainstorm.md`: Added **Recommended Breakdown Ordering** — 8-step idea→architect+UI flow (free collection → scope lock → coverage mapping → architect design → arch→UI sync → UI direction → completeness gate → save)
+- `workflows/brainstorm.md`: Added **Feature → Coverage Mapping** step — after scope lock, maps each Phase 1 feature to an architect page + UI screen; outputs `## Coverage` matrix in `notes.md`; warns on features with no coverage in either mode (non-blocking)
+- `workflows/brainstorm.md`: Added **Section 6C — Architect → UI Direction Sync** (`arch_to_ui_sync`) — reverse sync from Architect workspace to UI Direction; surfaces UI implications of architectural decisions (async flows, auth roles, error codes, constraints); `/sync-ui` command; records `## arch_to_ui_sync` in `notes.md`
+- `workflows/brainstorm.md`: Expanded pre-save completeness gate with **CHECK 4** — non-blocking warning when both workspaces are active but Phase 1 features have no coverage in either mode
+- `skills/vp-brainstorm/SKILL.md`: Breakdown loop documented (coverage matrix, `arch_to_ui_sync` reverse sync, completeness gate, recommended ordering)
+
+## [2.24.0] - 2026-04-18
+
+### Changed (ENH-060)
+- **`workflows/brainstorm.md`**: UI Direction Mode now proactively suggests itself when ≥1 UI keyword is found in the initial message (early-session detection), matching Architect Design Mode behavior — shows a 🎨 banner before topic selection
+- **`workflows/brainstorm.md`**: Lowered accumulation threshold from ≥3 to ≥1 signal; surface threshold lowered from ≥5 to ≥2 unique signals
+- **`workflows/brainstorm.md`**: Confirmation dialogue updated with 🎨 banner (was plain 💡), wording mirrors the Architect 🏗️ banner style
+- **`skills/vp-brainstorm/SKILL.md`**: Updated Background UI extraction docs — auto-suggestion, lower thresholds, 🎨 banner reference, ENH-060 tagged
+
 ## [2.23.0] - 2026-04-18
 
 ### Added

package/README.md

@@ -2,12 +2,12 @@
 
 **Autonomous Vibe Coding Framework / Bộ khung phát triển tự động có kiểm soát**
 
-[![Version](https://img.shields.io/badge/version-2.23.0-blue.svg)](CHANGELOG.md)
+[![Version](https://img.shields.io/badge/version-2.31.0-blue.svg)](CHANGELOG.md)
 [![License: MIT](https://img.shields.io/badge/License-MIT-green.svg)](LICENSE)
-[![Skills](https://img.shields.io/badge/skills-17-purple.svg)](#skills-reference)
+[![Skills](https://img.shields.io/badge/skills-18-purple.svg)](#skills-reference)
 [![Workflows](https://img.shields.io/badge/workflows-13-orange.svg)](#workflows)
 [![Templates](https://img.shields.io/badge/templates-17-cyan.svg)](#templates)
-[![Tests](https://img.shields.io/badge/tests-1140%20passing-brightgreen.svg)](tests/)
+[![Tests](https://img.shields.io/badge/tests-1324%20passing-brightgreen.svg)](tests/)
 [![GitHub](https://img.shields.io/github/stars/0-CODE/viepilot?style=social)](https://github.com/0-CODE/viepilot)
 
 **Versioning:** Shield **2.19.0** is the **ViePilot framework SemVer** tracked in `.viepilot/TRACKER.md` and `CHANGELOG.md`. The npm `package.json` field `version` (**2.19.0**) is the Node package identifier for this repo; use the framework version for milestone releases and docs.
@@ -29,7 +29,7 @@ Nếu ViePilot giúp ích cho bạn, bạn có thể ủng hộ một ly cafe:
 | Chỉ số / Metric | Giá trị / Value |
 |-----------------|-----------------|
 | Total LOC | **~51,713+** (`.md`, `.js`, `.cjs`, `.yml`, `.json`, `.sh`; không gồm `node_modules`) |
-| Skills | **17** |
+| Skills | **18** |
 | Workflows | **12** |
 | Templates | **17** (Project: 12, Phase: 5) |
 | CLI Commands | **18** (`vp-tools` 17 subcommands + `viepilot` installer) |
@@ -60,7 +60,7 @@ Tổng thể / Overall:  ██████████████████
 
 | Lĩnh vực / Area | Trạng thái | Chi tiết |
 |-----------------|------------|----------|
-| Core Skills (17) | ✅ Hoàn thiện | brainstorm, crystallize, auto, pause, resume, status, info, request, evolve, docs, update, task, debug, rollback, audit, ui-components, **proposal** |
+| Core Skills (18) | ✅ Hoàn thiện | brainstorm, crystallize, auto, pause, resume, status, info, request, evolve, docs, update, task, debug, rollback, audit, ui-components, proposal, **skills** |
 | Workflows (13) | ✅ Hoàn thiện | Full step-by-step guides với success criteria |
 | Project Templates (12) | ✅ Hoàn thiện | Placeholders cho customization (+ `VIEPILOT-META` FEAT-009) |
 | Phase Templates (5) | ✅ Hoàn thiện | Task tracking, verification, summary |
@@ -339,7 +339,7 @@ Full guide: `docs/dev/deployment.md`.
 
 ```
 viepilot/
-├── skills/                        # 17 Skill definitions
+├── skills/                        # 18 Skill definitions
 │   ├── vp-brainstorm/             # Thu thập ý tưởng
 │   │   └── SKILL.md
 │   ├── vp-crystallize/            # Tạo artifacts

package/bin/viepilot.cjs

@@ -33,6 +33,11 @@ ViePilot CLI
 Usage:
   viepilot install [options]
   viepilot uninstall [options]
+  viepilot scan-skills             Scan all adapter skill dirs and build skill registry
+  viepilot list-skills             List all indexed skills from registry
+  viepilot install-skill <source>  Install a third-party skill (npm:pkg, github:org/repo, ./path)
+  viepilot uninstall-skill <id>    Remove a skill from all adapter dirs
+  viepilot update-skill <id>       Re-install skill from its original source
   viepilot --help
   viepilot --list-targets
 
@@ -44,7 +49,7 @@ Install options:
   --help                     Show help
 
 Uninstall options:
-  --target <id|id,id|all>   Remove assets (claude-code: ~/.claude/skills/vp-*; cursor-*: ~/.cursor/skills/vp-*; antigravity: ~/.antigravity/skills/vp-*; codex: ~/.codex/skills/vp-*; copilot: ~/.config/gh-copilot/skills/vp-*)
+  --target <id|id,id|all>   Remove assets (claude-code: ~/.claude/skills/vp-*; cursor-*: ~/.cursor/skills/vp-*; antigravity: ~/.gemini/antigravity/skills/vp-*; codex: ~/.codex/skills/vp-*; copilot: ~/.config/gh-copilot/skills/vp-*)
   --yes                     Non-interactive mode (skip confirmations)
   --dry-run                 Print actions only, do not remove files
 `);
@@ -460,6 +465,108 @@ async function main() {
     printTargets();
     process.exit(0);
   }
+  if (command === 'scan-skills') {
+    const { scanSkills } = require('../lib/skill-registry.cjs');
+    const result = scanSkills();
+    const byAdapter = {};
+    for (const skill of result.skills) {
+      for (const adapter of skill.adapters) {
+        byAdapter[adapter] = (byAdapter[adapter] || 0) + 1;
+      }
+    }
+    console.log('Scanning skill directories...');
+    for (const p of result.scan_paths) {
+      const adapterName = Object.keys(byAdapter).find(a => p.includes(`.${a.replace('claude-code', 'claude')}`)) || '';
+      const count = byAdapter[adapterName];
+      const prefix = count != null ? '  ✓' : '  -';
+      console.log(`${prefix} ${p}${count != null ? `  — ${count} skill${count !== 1 ? 's' : ''} found` : '  — not installed (skipped)'}`);
+    }
+    const uniqueMulti = result.skills.filter(s => s.adapters.length > 1).length;
+    console.log(`\nRegistry written: ~/.viepilot/skill-registry.json`);
+    console.log(`Total: ${result.skills.length} unique skill${result.skills.length !== 1 ? 's' : ''}${uniqueMulti > 0 ? ` (${uniqueMulti} in multiple adapters)` : ''}`);
+    process.exit(0);
+  }
+
+  if (command === 'list-skills') {
+    const { loadRegistry } = require('../lib/skill-registry.cjs');
+    const registry = loadRegistry();
+    if (!registry) {
+      console.log('No registry found. Run vp-tools scan-skills first.');
+      process.exit(0);
+    }
+    console.log(`\nInstalled skills (from ~/.viepilot/skill-registry.json):`);
+    console.log(`Last scan: ${registry.last_scan}\n`);
+    if (registry.skills.length === 0) {
+      console.log('No skills indexed.');
+    } else {
+      const idW = Math.max(4, ...registry.skills.map(s => s.id.length));
+      const adW = Math.max(8, ...registry.skills.map(s => s.adapters.join(', ').length));
+      const capW = 30;
+      const header = `  ${'ID'.padEnd(idW)}  ${'Adapters'.padEnd(adW)}  Capabilities`;
+      const sep    = `  ${'─'.repeat(idW)}  ${'─'.repeat(adW)}  ${'─'.repeat(capW)}`;
+      console.log(header);
+      console.log(sep);
+      for (const skill of registry.skills) {
+        const caps = skill.capabilities.length > 0 ? skill.capabilities.join(', ') : '(legacy — no capabilities)';
+        console.log(`  ${skill.id.padEnd(idW)}  ${skill.adapters.join(', ').padEnd(adW)}  ${caps}`);
+      }
+    }
+    console.log('\nRun `vp-tools scan-skills` to refresh.');
+    process.exit(0);
+  }
+
+  if (command === 'install-skill') {
+    const source = rest[0];
+    if (!source) { console.error('Usage: vp-tools install-skill <source>'); process.exit(1); }
+    const { installSkill } = require('../lib/skill-installer.cjs');
+    const channel = source.startsWith('github:') ? 'github' : source.startsWith('./') || source.startsWith('/') ? 'local' : 'npm';
+    console.log(`Installing skill: ${source} (${channel})...`);
+    const result = await installSkill(source);
+    if (result.ok) {
+      for (const p of result.installedPaths) console.log(`  ✓ ${p}  — installed`);
+      console.log(`Done. Skill "${result.id}" is ready.`);
+      process.exit(0);
+    } else {
+      console.error(`Error: ${result.error}`);
+      process.exit(1);
+    }
+  }
+
+  if (command === 'uninstall-skill') {
+    const id = rest[0];
+    if (!id) { console.error('Usage: vp-tools uninstall-skill <id>'); process.exit(1); }
+    const { uninstallSkill } = require('../lib/skill-installer.cjs');
+    console.log(`Removing skill: ${id}...`);
+    const result = uninstallSkill(id);
+    if (result.ok) {
+      if (result.removedPaths.length === 0) {
+        console.log(`Skill "${id}" not found in any adapter dir.`);
+      } else {
+        for (const p of result.removedPaths) console.log(`  ✓ ${p}  — removed`);
+      }
+      process.exit(0);
+    } else {
+      console.error(`Error: ${result.error}`);
+      process.exit(1);
+    }
+  }
+
+  if (command === 'update-skill') {
+    const id = rest[0];
+    if (!id) { console.error('Usage: vp-tools update-skill <id>'); process.exit(1); }
+    const { updateSkill } = require('../lib/skill-installer.cjs');
+    console.log(`Updating skill: ${id}...`);
+    const result = await updateSkill(id);
+    if (result.ok) {
+      for (const p of result.installedPaths) console.log(`  ✓ ${p}  — updated`);
+      console.log(`Done. Skill "${result.id}" updated.`);
+      process.exit(0);
+    } else {
+      console.error(`Error: ${result.error}`);
+      process.exit(1);
+    }
+  }
+
   if (command !== 'install' && command !== 'uninstall') {
     console.error(`Unknown command: ${command}`);
     printHelp();

package/bin/vp-tools.cjs

@@ -1137,6 +1137,83 @@ ${colors.cyan}Examples:${colors.reset}
     process.exit(1);
   },
 
+  /**
+   * Output global skill registry as JSON (BUG-016 fix — shell-executable alternative to loadRegistry())
+   */
+  'get-registry': (args) => {
+    const { loadRegistry } = require('../lib/skill-registry.cjs');
+    const registry = loadRegistry();
+    if (!registry) {
+      process.stdout.write('null\n');
+      process.exit(0);
+    }
+    const idFilter = args[0] === '--id' ? args[1] : null;
+    if (idFilter) {
+      const skill = (registry.skills || []).find(s => s.id === idFilter) || null;
+      process.stdout.write(JSON.stringify(skill) + '\n');
+    } else {
+      process.stdout.write(JSON.stringify(registry) + '\n');
+    }
+    process.exit(0);
+  },
+
+  /**
+   * ENH-072: Check npm registry for newer ViePilot version (24h cached).
+   * --silent  exit 0 if up-to-date (no output); exit 1 + print latest if update available
+   * --json    print { installed, latest, has_update }; always exit 0
+   * --force   bypass 24h cache
+   */
+  'check-update': (args) => {
+    const silent = args.includes('--silent');
+    const json = args.includes('--json');
+    const force = args.includes('--force');
+    const { checkLatestVersion } = require('../lib/viepilot-update.cjs');
+
+    checkLatestVersion({ force })
+      .then(({ upToDate, installed, latest }) => {
+        const has_update = !upToDate;
+        if (json) {
+          process.stdout.write(JSON.stringify({ installed, latest, has_update }) + '\n');
+          process.exit(0);
+        }
+        if (silent) {
+          if (has_update) {
+            process.stdout.write(latest + '\n');
+            process.exit(1);
+          } else {
+            process.exit(0);
+          }
+        }
+        // default human-readable output
+        if (has_update) {
+          process.stdout.write(
+            `\n┌──────────────────────────────────────────────────────────────────┐\n` +
+            `│ ✨ ViePilot v${latest} available  (installed: v${installed})`.padEnd(69) + '│\n' +
+            `│    npm i -g viepilot && vp-tools install`.padEnd(69) + '│\n' +
+            `└──────────────────────────────────────────────────────────────────┘\n`
+          );
+          process.exit(1);
+        } else {
+          process.stdout.write(`ViePilot ${installed} is up to date.\n`);
+          process.exit(0);
+        }
+      })
+      .catch(() => {
+        process.exit(0);
+      });
+  },
+
+  /**
+   * Scan installed skills and rebuild ~/.viepilot/skill-registry.json (BUG-019)
+   */
+  'scan-skills': (_args) => {
+    const { scanSkills } = require('../lib/skill-registry.cjs');
+    const result = scanSkills();
+    const count = (result && result.skills) ? result.skills.length : 0;
+    process.stdout.write(`✔ Scanned ${count} skill${count !== 1 ? 's' : ''} → ~/.viepilot/skill-registry.json\n`);
+    process.exit(0);
+  },
+
   /**
    * Help
    */
@@ -1224,6 +1301,35 @@ ${colors.cyan}Examples:${colors.reset}
         options: ['--json: Machine-readable JSON output'],
         examples: ['vp-tools info', 'vp-tools info --json'],
       },
+      'get-registry': {
+        usage: 'vp-tools get-registry [--id <skill-id>]',
+        description: 'Output global skill registry as JSON (enables workflow shell integration)',
+        options: ['--id <id>: Output only the matching skill object (null if not found)'],
+        examples: [
+          'vp-tools get-registry',
+          'vp-tools get-registry --id frontend-design',
+        ],
+      },
+      'scan-skills': {
+        usage: 'vp-tools scan-skills',
+        description: 'Scan installed skills and rebuild ~/.viepilot/skill-registry.json',
+        options: [],
+        examples: ['vp-tools scan-skills'],
+      },
+      'check-update': {
+        usage: 'vp-tools check-update [--silent] [--json] [--force]',
+        description: 'Check for latest ViePilot version on npm registry (24h cached)',
+        options: [
+          '--silent  Exit 0 if up-to-date (no output); exit 1 + print latest if update available',
+          '--json    Print { installed, latest, has_update }; always exit 0',
+          '--force   Bypass 24h cache and check now',
+        ],
+        examples: [
+          'vp-tools check-update --silent',
+          'vp-tools check-update --json',
+          'vp-tools check-update --force',
+        ],
+      },
       update: {
         usage: 'vp-tools update [--dry-run] [--yes] [--global]',
         description: 'Update viepilot to npm latest (local dependency, global install, or explicit --global)',
@@ -1282,6 +1388,9 @@ ${colors.cyan}Commands:${colors.reset}
   ${colors.bold}hooks${colors.reset} scaffold|install [--adapter] scaffold: print snippet; install: write to settings.json
   ${colors.bold}config${colors.reset} <get|set|reset>    Read/write language config (~/.viepilot/config.json)
   ${colors.bold}save-state${colors.reset}               Save current state for precise resume
+  ${colors.bold}get-registry${colors.reset} [--id <id>] Output global skill registry as JSON
+  ${colors.bold}scan-skills${colors.reset}               Scan installed skills → ~/.viepilot/skill-registry.json
+  ${colors.bold}check-update${colors.reset} [--silent]   Check for latest ViePilot version on npm (24h cached)
   ${colors.bold}help${colors.reset} [command]           Show help (optionally for specific command)
 
 ${colors.cyan}Examples:${colors.reset}