@arthai/agents 1.0.0 → 1.0.1

package/README.md

@@ -1,123 +1,126 @@
-# Claude Agents
+# @arthai/agents
 
-A portable toolkit of 18 AI agents, 18 skills, and 13 hooks for [Claude Code](https://docs.anthropic.com/en/docs/claude-code). Install once at `~/.claude-agents/`, symlink into any project — all projects stay in sync automatically.
+AI development toolkit for Claude Code — agents, skills, and hooks organized into installable bundles.
 
-## Quick Start
+## Install
 
 ```bash
-# 1. Clone + license
-git clone git@github.com:ArthTech-AI/claude-agents.git ~/.claude-agents
-~/.claude-agents/install.sh --key ARTH-XXXX-XXXX-XXXX-XXXX
+# Install a bundle into your project
+npx @arthai/agents install forge .
 
-# 2. Install into your project
-~/.claude-agents/install.sh /path/to/project
+# Install multiple bundles
+npx @arthai/agents install forge spark shield .
+
+# See all available bundles
+npx @arthai/agents list
+
+# Show what's in a bundle
+npx @arthai/agents info forge
 ```
 
-Open Claude Code — agents, skills, and hooks are active. Auto-syncs on every session start.
-
-First session: type `/calibrate` to deep-learn your project (reads source code, recommends agents, skills, MCP servers).
-
-### Other Install Modes
-
-| Command | When to use |
-|---------|-------------|
-| `install.sh --assess .` | See what you have vs the toolkit (read-only) |
-| `install.sh --setup .` | Change category selections |
-| `install.sh --upgrade .` | Upgrade existing project (converts stale files) |
-| `install.sh --convert .` | Migrate from copy-based install |
-| `install.sh --status .` | Show what's linked/overridden/missing |
-
-## What You Get
-
-**18 Agents** — specialized AI workers Claude delegates to automatically:
-
-| Tier | Model | Cost | Agents |
-|------|-------|------|--------|
-| Cheap | Haiku | 1x | explore-light, ops, qa-baseline-updater, qa-test-promoter |
-| Mid | Sonnet | 10x | frontend, python-backend, sre, qa, qa-domain, qa-e2e, qa-challenger, code-reviewer, design-studio:think, design-studio:critique |
-| Expensive | Opus | 60x | architect, product-manager, gtm-expert, design-studio:create |
-
-**19 Skills** — invoke with `/name` in Claude Code:
-
-| Skill | What it does |
-|-------|-------------|
-| `/calibrate` | Deep-learn project, recommend + install MCP servers, agents, skills, workflows, knowledge base |
-| `/onboard` | Session start — project briefing + work prioritization |
-| `/planning <feature>` | Spawn planning team (PM + Architect + optional Design/GTM) |
-| `/implement <feature>` | Spawn implementation team from plan |
-| `/pr` | QA + PR creation + post-merge cleanup |
-| `/qa [full\|staging\|prod]` | Run QA checks at different levels |
-| `/ci-fix [ci\|staging\|prod]` | Auto-remediate CI/CD failures (3 retries) |
-| `/issue <title>` | GitHub issue management |
-| `/explore <query>` | Fast codebase search (Haiku) |
-| `/scan` | Auto-populate CLAUDE.md from codebase |
-| `/sync` | Sync toolkit, check status, contribute back |
-| `/restart` | Kill and restart local dev servers |
-| `/sre <cmd>` | SRE operations: status, health, logs, debug |
-| `/qa-incident` | Log a QA incident |
-| `/qa-learn` | Review QA knowledge base |
-| `/discord-ops` | Discord ChatOps reference |
-| `/superpowers` | Advanced operations |
-| `/autopilot [issue]` | Autonomous work loop — assess, plan, implement, QA, PR |
-| `/railway` | Railway-specific ops (needs `--railway` flag) |
-
-**13 Hooks** — run automatically on Claude Code events:
-
-| Hook | Event | Purpose |
-|------|-------|---------|
-| triage-router.sh | UserPromptSubmit | Cost-aware routing to cheaper agents |
-| session-bootstrap.sh | SessionStart | Compact project briefing |
-| sync-agents.sh | SessionStart | Auto-pull repo + re-sync |
-| pre-bash-guard.sh | PreToolUse | Blocks dangerous commands |
-| pre-edit-guard.sh | PreToolUse | Prevents edits to protected files |
-| pre-task-context.sh | PreToolUse | Injects CLAUDE.md into spawned agents |
-| post-test-summary.sh | PostToolUse | Summarizes test/lint output |
-| post-deploy-health.sh | PostToolUse | Health checks after deploys |
-| post-edit-lint.sh | PostToolUse | Lint check after file edits |
-| sync-worktree.sh | PostToolUse | Syncs config across worktrees |
-| sync-railway-worktree.sh | PostToolUse | Railway worktree sync |
-| post-git-state.sh | PostToolUse | Detects merges, resets autopilot phase |
-| session-start.sh | SessionStart | Legacy session start |
-
-## Install Modes
-
-| Command | Purpose |
-|---------|---------|
-| `install.sh .` | Sync — create missing symlinks, skip existing files |
-| `install.sh --init .` | Greenfield — scaffold dirs + CLAUDE.md + settings.json + sync |
-| `install.sh --setup .` | Interactive — pick which categories to install |
-| `install.sh --convert .` | Migrate — replace matching copies with symlinks |
-| `install.sh --assess .` | Brownfield analysis — read-only, shows diffs |
-| `install.sh --upgrade .` | Brownfield upgrade — interactive per-file conversion |
-| `install.sh --status .` | Status report — what's linked, overridden, missing |
-| `install.sh --uninstall .` | Clean removal — restore backups, remove symlinks |
-| `install.sh --generate-config .` | Generate config from existing .claude/ contents |
-| `install.sh --key KEY` | Store license key |
-
-Flags: `--railway` (Railway skills), `--yes` (auto-approve upgrades), `--skip file1,file2` (exclude files)
-
-## How It Works
-
-Symlinks from your project to `~/.claude-agents/`. Update the repo, all projects update.
+## Bundles
+
+Pick what you need. Each bundle is a self-contained set of agents, skills, and hooks.
+
+### Development
+
+| Bundle | Command | What you get |
+|--------|---------|-------------|
+| **forge** | `npx @arthai/agents install forge .` | Full dev workflow: `/planning`, `/implement`, `/qa`, `/pr`, `/precheck`, `/review-pr` + 6 agents (architect, PM, backend, frontend, QA, code-reviewer) + triage router hook |
+| **scalpel** | `npx @arthai/agents install scalpel .` | Bug fixing: `/fix`, `/ci-fix`, `/issue` + code-reviewer agent + edit guard hook |
+| **spark** | `npx @arthai/agents install spark .` | Project setup: `/onboard`, `/calibrate`, `/scan`, `/setup` + setup and explore agents |
+
+### Operations
+
+| Bundle | Command | What you get |
+|--------|---------|-------------|
+| **sentinel** | `npx @arthai/agents install sentinel .` | SRE + ops: `/sre`, `/incident`, `/restart` + SRE and ops agents + deploy health + git state hooks |
+| **prism** | `npx @arthai/agents install prism .` | Deep QA: `/qa`, `/qa-learn`, `/qa-incident` + 6 QA agents (qa, e2e, domain, challenger, baseline, promoter) |
+| **shield** | `npx @arthai/agents install shield .` | Safety guardrails: bash guard, edit guard, session bootstrap, triage router hooks (no skills — hooks only) |
+
+### Strategy
+
+| Bundle | Command | What you get |
+|--------|---------|-------------|
+| **canvas** | `npx @arthai/agents install canvas .` | Design: `/planning` + design-studio agents (think, create, critique) + frontend agent |
+| **compass** | `npx @arthai/agents install compass .` | Product: `/planning` + PM, GTM, user researcher, content strategist agents |
+| **counsel** | `npx @arthai/agents install counsel .` | Consulting: 13 skills (client-discovery, proposals, deliverables, ROI, etc.) + 4 consulting agents |
+
+### Power
 
+| Bundle | Command | What you get |
+|--------|---------|-------------|
+| **cruise** | `npx @arthai/agents install cruise .` | Autonomous mode: `/autopilot` — picks up issues, implements, QAs, creates PRs. Auto-installs forge + scalpel + sentinel. |
+| **prime** | `npx @arthai/agents install prime .` | Everything. All 25 agents, 37 skills, 16 hooks. |
+
+## Workflows
+
+After installing, use skills in Claude Code:
+
+**Start a new project:**
 ```
-your-project/.claude/agents/architect.md  →  ~/.claude-agents/agents/architect.md
+/onboard                    # get project briefing
+/calibrate                  # auto-configure toolkit for your codebase
+/planning my-feature        # design the feature with architect + PM agents
+/implement my-feature       # build with parallel agent team
 ```
 
-**Override**: Replace any symlink with a regular file. Installer skips it forever.
-**Auto-sync**: `sync-agents.sh` hook pulls latest on every session start.
-**Auto-adapt**: `/calibrate` reads actual source code (not just config files) to tailor agents, skills, MCP servers, and workflows to your project.
-**Cost routing**: Triage router delegates to Haiku/Sonnet before using Opus.
+**Fix a bug:**
+```
+/fix #42                    # root cause analysis + fix + regression tests
+/ci-fix                     # auto-repair CI failures
+```
 
-## Docs
+**Ship code:**
+```
+/precheck                   # local CI in 30s
+/qa commit                  # targeted QA on changed files
+/pr                         # create PR with QA results
+```
 
-- **[GETTING-STARTED.md](GETTING-STARTED.md)** — Full user guide (install, usage, customization, contributing)
-- **[tests/README.md](tests/README.md)** — Test suite quick start
-- **[tests/TESTING.md](tests/TESTING.md)** — Full test guide (suites, fixtures, libraries, writing tests)
+**Autonomous mode:**
+```
+/autopilot                  # picks up issues → implements → QAs → creates PRs
+```
+
+**Operations:**
+```
+/sre status                 # health check all services
+/incident                   # triage + parallel diagnosis
+/restart                    # restart local dev servers
+```
+
+## Uninstall
+
+```bash
+npx @arthai/agents uninstall forge .
+```
+
+Remove everything:
+
+```bash
+rm -rf .claude/skills .claude/agents .claude/hooks
+```
+
+Note: hooks merged into `.claude/settings.json` need to be removed manually.
+
+## Troubleshooting
+
+| Problem | Fix |
+|---------|-----|
+| Skills not showing | Restart Claude Code in the project |
+| `/planning` says unknown skill | Verify files exist: `ls .claude/skills/planning/SKILL.md` |
+| Install says "not found in dist/" | Update: `npx @arthai/agents@latest install forge .` |
+
+## Also available as Claude Code plugin
+
+```
+/plugin marketplace add ArthTech-AI/arthai-marketplace
+/plugin install forge@arthai-marketplace
+```
 
 ## Requirements
 
 - [Claude Code CLI](https://docs.anthropic.com/en/docs/claude-code)
-- `CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1` in settings
+- Node.js 18+
 - License key (format: `ARTH-XXXX-XXXX-XXXX-XXXX`)
-- (Optional) [Railway CLI](https://docs.railway.app/guides/cli) for Railway skills

package/VERSION

@@ -1 +1 @@
-1.0.0
+1.0.1

package/bin/cli.js

@@ -26,6 +26,10 @@ switch (command) {
   case 'info':
     handleInfo(args[1]);
     break;
+  case 'uninstall':
+  case 'remove':
+    handleUninstall(args.slice(1));
+    break;
   case 'legacy':
     handleLegacy(args.slice(1));
     break;
@@ -94,7 +98,13 @@ function handleInstall(rawArgs) {
     process.exit(1);
   }
 
-  const pluginsTargetDir = path.join(targetDir, '.claude', 'plugins');
+  const claudeDir = path.join(targetDir, '.claude');
+  const skillsDir = path.join(claudeDir, 'skills');
+  const agentsDir = path.join(claudeDir, 'agents');
+
+  let totalCommands = 0;
+  let totalAgents = 0;
+  let totalHooks = 0;
 
   for (const name of resolved) {
     // Defense-in-depth: verify the resolved path stays inside PLUGINS_DIR
@@ -107,11 +117,13 @@ function handleInstall(rawArgs) {
       console.error(`Plugin '${name}' not found in dist/plugins/. Run compiler.sh first.`);
       process.exit(1);
     }
-    copyPlugin(pluginDir, pluginsTargetDir, name);
-    console.log(`  Installed: ${name} → ${path.join(pluginsTargetDir, name)}`);
+    const counts = installBundle(pluginDir, claudeDir, skillsDir, agentsDir, name);
+    totalCommands += counts.commands;
+    totalAgents += counts.agents;
+    totalHooks += counts.hooks;
   }
 
-  console.log(`\nDone. ${resolved.length} plugin(s) installed to ${pluginsTargetDir}`);
+  console.log(`\nDone. ${resolved.length} bundle(s) installed: ${totalCommands} skills, ${totalAgents} agents, ${totalHooks} hooks`);
 }
 
 // ── list ──────────────────────────────────────────────────────────────────────
@@ -199,6 +211,71 @@ function handleLegacy(legacyArgs) {
   process.exit(result.status || 0);
 }
 
+// ── uninstall ────────────────────────────────────────────────────────────────
+
+function handleUninstall(rawArgs) {
+  if (rawArgs.length === 0) {
+    console.error('Usage: arthai uninstall <bundle> [bundle...] [target-path]');
+    process.exit(1);
+  }
+
+  const targetDir = looksLikePath(rawArgs[rawArgs.length - 1])
+    ? rawArgs.pop()
+    : process.cwd();
+
+  const bundleNames = rawArgs.filter(a => isValidBundleName(a));
+  if (bundleNames.length === 0) {
+    console.error('No valid bundle names provided.');
+    process.exit(1);
+  }
+
+  const claudeDir = path.join(targetDir, '.claude');
+  const skillsDir = path.join(claudeDir, 'skills');
+  const agentsDir = path.join(claudeDir, 'agents');
+
+  for (const name of bundleNames) {
+    const pluginDir = path.resolve(PLUGINS_DIR, name);
+    if (!fs.existsSync(pluginDir)) {
+      console.error(`Bundle '${name}' not found in dist/plugins/.`);
+      continue;
+    }
+
+    let removed = 0;
+
+    // Remove skills that came from this bundle's commands/
+    const commandsDir = path.join(pluginDir, 'commands');
+    if (fs.existsSync(commandsDir)) {
+      for (const file of fs.readdirSync(commandsDir)) {
+        if (!file.endsWith('.md')) continue;
+        const skillName = file.replace('.md', '');
+        const skillPath = path.join(skillsDir, skillName);
+        if (fs.existsSync(skillPath)) {
+          fs.rmSync(skillPath, { recursive: true });
+          removed++;
+        }
+      }
+    }
+
+    // Remove agents that came from this bundle
+    const srcAgents = path.join(pluginDir, 'agents');
+    if (fs.existsSync(srcAgents)) {
+      for (const file of fs.readdirSync(srcAgents)) {
+        if (!file.endsWith('.md')) continue;
+        const agentPath = path.join(agentsDir, file);
+        if (fs.existsSync(agentPath)) {
+          fs.unlinkSync(agentPath);
+          removed++;
+        }
+      }
+    }
+
+    console.log(`  Removed ${name}: ${removed} files`);
+  }
+
+  console.log('\nNote: hooks in .claude/settings.json were not removed — edit manually if needed.');
+  console.log('Done.');
+}
+
 // ── helpers ───────────────────────────────────────────────────────────────────
 
 // Validate bundle names: only lowercase alphanumeric and hyphens, max 64 chars.
@@ -257,17 +334,76 @@ function resolveDependencies(names) {
   return resolved;
 }
 
-function copyPlugin(sourceDir, targetPluginsDir, name) {
-  const dest = path.join(targetPluginsDir, name);
+function installBundle(pluginDir, claudeDir, skillsDir, agentsDir, name) {
+  const counts = { commands: 0, agents: 0, hooks: 0 };
+
+  // 1. Copy commands/ → .claude/skills/<name>/SKILL.md (each .md becomes a skill dir)
+  const commandsDir = path.join(pluginDir, 'commands');
+  if (fs.existsSync(commandsDir)) {
+    for (const file of fs.readdirSync(commandsDir)) {
+      if (!file.endsWith('.md')) continue;
+      const skillName = file.replace('.md', '');
+      const destDir = path.join(skillsDir, skillName);
+      fs.mkdirSync(destDir, { recursive: true });
+      fs.copyFileSync(path.join(commandsDir, file), path.join(destDir, 'SKILL.md'));
+      counts.commands++;
+    }
+    console.log(`  ${name}: ${counts.commands} skills → ${skillsDir}`);
+  }
+
+  // 2. Copy agents/ → .claude/agents/
+  const srcAgents = path.join(pluginDir, 'agents');
+  if (fs.existsSync(srcAgents)) {
+    fs.mkdirSync(agentsDir, { recursive: true });
+    for (const file of fs.readdirSync(srcAgents)) {
+      if (!file.endsWith('.md')) continue;
+      fs.copyFileSync(path.join(srcAgents, file), path.join(agentsDir, file));
+      counts.agents++;
+    }
+    console.log(`  ${name}: ${counts.agents} agents → ${agentsDir}`);
+  }
+
+  // 3. Merge hooks into .claude/settings.json
+  const hooksJson = path.join(pluginDir, 'hooks', 'hooks.json');
+  if (fs.existsSync(hooksJson)) {
+    const pluginHooks = JSON.parse(fs.readFileSync(hooksJson, 'utf8'));
+    const settingsPath = path.join(claudeDir, 'settings.json');
+    let settings = {};
+    if (fs.existsSync(settingsPath)) {
+      try { settings = JSON.parse(fs.readFileSync(settingsPath, 'utf8')); } catch (e) { settings = {}; }
+    }
+    if (!settings.hooks) settings.hooks = {};
+
+    // Copy hook scripts to .claude/hooks/
+    const hooksDir = path.join(claudeDir, 'hooks');
+    fs.mkdirSync(hooksDir, { recursive: true });
+    const srcHooksDir = path.join(pluginDir, 'hooks');
+    for (const file of fs.readdirSync(srcHooksDir)) {
+      if (!file.endsWith('.sh')) continue;
+      const dest = path.join(hooksDir, file);
+      fs.copyFileSync(path.join(srcHooksDir, file), dest);
+      fs.chmodSync(dest, 0o755);
+    }
+
+    // Merge hook entries (rewrite CLAUDE_PLUGIN_ROOT to project hooks path)
+    const hookData = pluginHooks.hooks || pluginHooks;
+    for (const [event, entries] of Object.entries(hookData)) {
+      if (!settings.hooks[event]) settings.hooks[event] = [];
+      for (const entry of entries) {
+        // Rewrite command paths from ${CLAUDE_PLUGIN_ROOT} to $CLAUDE_PROJECT_DIR/.claude
+        const rewritten = JSON.parse(
+          JSON.stringify(entry).replace(/\$\{CLAUDE_PLUGIN_ROOT\}/g, '$CLAUDE_PROJECT_DIR/.claude')
+        );
+        settings.hooks[event].push(rewritten);
+        counts.hooks++;
+      }
+    }
 
-  // Remove existing
-  if (fs.existsSync(dest)) {
-    fs.rmSync(dest, { recursive: true });
+    fs.writeFileSync(settingsPath, JSON.stringify(settings, null, 2) + '\n');
+    console.log(`  ${name}: ${counts.hooks} hook entries merged into settings.json`);
   }
-  fs.mkdirSync(dest, { recursive: true });
 
-  // Recursive copy
-  copyDirRecursive(sourceDir, dest);
+  return counts;
 }
 
 function copyDirRecursive(src, dest) {

package/compiler.sh

@@ -153,6 +153,7 @@ for bundle_file in "${BUNDLE_FILES[@]}"; do
     mkdir -p "$PLUGIN_DIR/.claude-plugin"
     mkdir -p "$PLUGIN_DIR/agents"
     mkdir -p "$PLUGIN_DIR/skills"
+    mkdir -p "$PLUGIN_DIR/commands"
     mkdir -p "$PLUGIN_DIR/hooks"
 
     # ── 1. Write plugin.json ──────────────────────────────────────────────────
@@ -163,8 +164,7 @@ for bundle_file in "${BUNDLE_FILES[@]}"; do
         --arg name "$NAME" \
         --arg description "$DESC" \
         --arg version "$VERSION" \
-        --arg author "Arth AI" \
-        '{"name": $name, "description": $description, "version": $version, "author": $author}' \
+        '{"name": $name, "description": $description, "version": $version, "author": {"name": "Arth AI"}}' \
         > "$PLUGIN_DIR/.claude-plugin/plugin.json"
 
     # ── 2. Determine agents/skills/hooks lists ────────────────────────────────
@@ -223,19 +223,73 @@ for bundle_file in "${BUNDLE_FILES[@]}"; do
         ((AGENT_COUNT++)) || true
     done
 
-    # ── 4. Copy skills ────────────────────────────────────────────────────────
+    # ── 4. Copy skills → commands/ (user-invocable) or skills/ (agent-invoked) ──
+    #
+    # Claude Code plugin system:
+    #   commands/*.md  → slash commands users type (/plugin:name)
+    #   skills/*/SKILL.md → context skills Claude auto-invokes
+    #
+    # We detect user-invocable skills by checking for "user-invocable: true"
+    # in SKILL.md frontmatter. These become flat .md files in commands/.
+    # Custom frontmatter fields (user-invocable, arguments) are stripped
+    # since Claude Code doesn't recognize them.
 
     SKILL_COUNT=0
+    CMD_COUNT=0
     for skill_path in "${SKILL_PATHS[@]:-}"; do
         [ -z "$skill_path" ] && continue
         src="$SCRIPT_DIR/$skill_path"
-        if [ -d "$src" ]; then
-            # Use -P to avoid following symlinks (prevents cycles from install.sh artifacts)
-            cp -rP "$src" "$PLUGIN_DIR/skills/"
-            ((SKILL_COUNT++)) || true
+        if [ -d "$src" ] && [ -f "$src/SKILL.md" ]; then
+            # Check if this is a user-invocable skill → goes to commands/
+            if grep -q 'user-invocable:.*true' "$src/SKILL.md" 2>/dev/null; then
+                # Copy SKILL.md to commands/<name>.md, stripping custom fields
+                skill_name=$(basename "$src")
+                sed '/^user-invocable:/d; /^arguments:/d' "$src/SKILL.md" \
+                    > "$PLUGIN_DIR/commands/$skill_name.md"
+                # Copy any supporting files (references, scripts) alongside
+                find "$src" -maxdepth 1 -not -name SKILL.md -not -name "$(basename "$src")" -not -type l -not -path "$src" | while IFS= read -r extra; do
+                    if [ -d "$extra" ]; then
+                        rsync -a --no-links "$extra" "$PLUGIN_DIR/commands/"
+                    elif [ -f "$extra" ]; then
+                        cp "$extra" "$PLUGIN_DIR/commands/"
+                    fi
+                done
+                ((CMD_COUNT++)) || true
+            else
+                # Agent-invoked skill → stays in skills/
+                rsync -a --no-links "$src" "$PLUGIN_DIR/skills/"
+                ((SKILL_COUNT++)) || true
+            fi
+        elif [ -d "$src" ]; then
+            # Directory without SKILL.md — check for nested skills (container dir)
+            if find "$src" -mindepth 2 -name "SKILL.md" -print -quit 2>/dev/null | grep -q .; then
+                # Container dir — process each sub-skill
+                for sub_skill in "$src"/*/; do
+                    [ -f "$sub_skill/SKILL.md" ] || continue
+                    if grep -q 'user-invocable:.*true' "$sub_skill/SKILL.md" 2>/dev/null; then
+                        sub_name=$(basename "$sub_skill")
+                        sed '/^user-invocable:/d; /^arguments:/d' "$sub_skill/SKILL.md" \
+                            > "$PLUGIN_DIR/commands/$sub_name.md"
+                        ((CMD_COUNT++)) || true
+                    else
+                        rsync -a --no-links "$sub_skill" "$PLUGIN_DIR/skills/"
+                        ((SKILL_COUNT++)) || true
+                    fi
+                done
+            else
+                rsync -a --no-links "$src" "$PLUGIN_DIR/skills/"
+                ((SKILL_COUNT++)) || true
+            fi
         elif [ -f "$src" ]; then
-            cp "$src" "$PLUGIN_DIR/skills/"
-            ((SKILL_COUNT++)) || true
+            # Flat .md file (e.g., explore.md)
+            if grep -q 'user-invocable:.*true' "$src" 2>/dev/null; then
+                sed '/^user-invocable:/d; /^arguments:/d' "$src" \
+                    > "$PLUGIN_DIR/commands/$(basename "$src")"
+                ((CMD_COUNT++)) || true
+            else
+                cp "$src" "$PLUGIN_DIR/skills/"
+                ((SKILL_COUNT++)) || true
+            fi
         else
             echo "  WARNING: Skill not found: $src" >&2
             ((ERRORS++)) || true
@@ -292,7 +346,12 @@ for bundle_file in "${BUNDLE_FILES[@]}"; do
         echo "$HOOKS_JSON" | jq '.' > "$PLUGIN_DIR/hooks/hooks.json"
     fi
 
-    echo "  agents=$AGENT_COUNT  skills=$SKILL_COUNT  hooks=$HOOK_COUNT"
+    # Remove empty directories
+    rmdir "$PLUGIN_DIR/commands" 2>/dev/null || true
+    rmdir "$PLUGIN_DIR/skills" 2>/dev/null || true
+    rmdir "$PLUGIN_DIR/hooks" 2>/dev/null || true
+
+    echo "  agents=$AGENT_COUNT  commands=$CMD_COUNT  skills=$SKILL_COUNT  hooks=$HOOK_COUNT"
     ((COMPILED++)) || true
 done
 

package/dist/plugins/canvas/.claude-plugin/plugin.json

@@ -1,6 +1,8 @@
 {
   "name": "canvas",
   "description": "Design-driven development — adversarial design critique + frontend",
-  "version": "1.0.0",
-  "author": "Arth AI"
+  "version": "1.0.1",
+  "author": {
+    "name": "Arth AI"
+  }
 }

package/dist/plugins/canvas/{skills/planning/SKILL.md → commands/planning.md}

@@ -1,8 +1,6 @@
 ---
 name: planning
 description: "Spin up an adversarial planning team for a feature. Usage: /planning <feature-name> [--design] [--gtm] [--fast] [--lite] [-- <brief>]"
-user-invocable: true
-arguments: "<feature-name> <--design> <--gtm> <--fast> <--lite> <-- brief>"
 ---
 
 # Planning Team Skill

package/dist/plugins/compass/.claude-plugin/plugin.json

@@ -1,6 +1,8 @@
 {
   "name": "compass",
   "description": "Product strategy — PM, GTM, user research, content strategy",
-  "version": "1.0.0",
-  "author": "Arth AI"
+  "version": "1.0.1",
+  "author": {
+    "name": "Arth AI"
+  }
 }

package/dist/plugins/compass/{skills/planning/SKILL.md → commands/planning.md}

@@ -1,8 +1,6 @@
 ---
 name: planning
 description: "Spin up an adversarial planning team for a feature. Usage: /planning <feature-name> [--design] [--gtm] [--fast] [--lite] [-- <brief>]"
-user-invocable: true
-arguments: "<feature-name> <--design> <--gtm> <--fast> <--lite> <-- brief>"
 ---
 
 # Planning Team Skill

package/dist/plugins/counsel/.claude-plugin/plugin.json

@@ -1,6 +1,8 @@
 {
   "name": "counsel",
   "description": "AI consulting toolkit — client discovery, proposals, deliverables",
-  "version": "1.0.0",
-  "author": "Arth AI"
+  "version": "1.0.1",
+  "author": {
+    "name": "Arth AI"
+  }
 }

package/dist/plugins/counsel/{skills/client-discovery/SKILL.md → commands/client-discovery.md}

@@ -1,8 +1,6 @@
 ---
 name: client-discovery
 description: "Client discovery and AI readiness assessment. Usage: /client-discovery <client-name> [-- <brief>]"
-user-invocable: true
-arguments: "<client-name> [-- brief]"
 ---
 
 # Client Discovery & AI Readiness Assessment

package/dist/plugins/{prime/skills/consulting/SKILL.md → counsel/commands/consulting.md}

@@ -1,8 +1,6 @@
 ---
 name: consulting
 description: "Run a full AI consulting engagement. Usage: /consulting <client-name> [--phase discovery|assess|propose|design|deliver|track] [-- <brief>]"
-user-invocable: true
-arguments: "<client-name> [--phase] [-- brief]"
 ---
 
 # AI Consulting Engagement Orchestrator

package/dist/plugins/{prime/skills/deliverable-builder/SKILL.md → counsel/commands/deliverable-builder.md}

@@ -1,8 +1,6 @@
 ---
 name: deliverable-builder
 description: "Build client-ready consulting deliverables. Usage: /deliverable-builder <client-name> [--type final-report|status-update|board-deck|implementation-guide|training-plan|change-management|data-strategy]"
-user-invocable: true
-arguments: "<client-name> [--type]"
 ---
 
 # Deliverable Builder Skill

package/dist/plugins/counsel/{skills/engagement-tracker/SKILL.md → commands/engagement-tracker.md}

@@ -1,8 +1,6 @@
 ---
 name: engagement-tracker
 description: "Track consulting engagement progress and milestones. Usage: /engagement-tracker <client-name> [--action status|update|milestone]"
-user-invocable: true
-arguments: "<client-name> [--action status|update|milestone]"
 ---
 
 # Engagement Tracker Skill

package/dist/plugins/{prime/skills/market-research/SKILL.md → counsel/commands/market-research.md}

@@ -1,8 +1,6 @@
 ---
 name: market-research
 description: "Research industry AI adoption, competitors, and trends. Usage: /market-research <client-name> [--focus industry|competitors|trends]"
-user-invocable: true
-arguments: "<client-name> [--focus industry|competitors|trends]"
 ---
 
 # Market Research Skill