@ekkos/cli 0.2.0

package/templates/claude-plugins-admin/autonomous-admin-agent/commands/agent.md

@@ -0,0 +1,595 @@
+# Autonomous Admin Agent
+
+**ADMIN ONLY** - Intelligent agent that monitors ekkOS infrastructure 24/7 and automatically uses admin plugins to maintain system health.
+
+## Overview
+
+The Autonomous Admin Agent is your AI DevOps engineer. It continuously monitors all ekkOS services and intelligently decides when to use admin plugins to:
+- Prevent issues before they happen
+- Fix problems automatically
+- Alert you only when human intervention is needed
+- Optimize performance proactively
+
+## Architecture
+
+```
+┌─────────────────────────────────────────────────────┐
+│          Autonomous Admin Agent                     │
+│                                                     │
+│  🧠 Intelligent Decision Making                    │
+│  ├─ Monitor all services continuously              │
+│  ├─ Detect anomalies and issues                    │
+│  ├─ Decide which plugin to use                     │
+│  ├─ Take preventive actions                        │
+│  └─ Alert admin when needed                        │
+└──────────────────┬──────────────────────────────────┘
+                   │
+         Uses admin plugins as tools
+                   │
+    ┌──────────────┼──────────────┬────────────┐
+    │              │              │            │
+    ▼              ▼              ▼            ▼
+┌─────────┐  ┌──────────┐  ┌──────────┐  ┌────────┐
+│ Deploy  │  │  Memory  │  │   MCP    │  │Database│
+│Guardian │  │ Monitor  │  │ Manager  │  │Manager │
+└─────────┘  └──────────┘  └──────────┘  └────────┘
+```
+
+## Commands
+
+### `/agent start`
+
+Start the autonomous agent to monitor infrastructure.
+
+```bash
+# Start monitoring
+/agent start
+
+# Start with specific focus
+/agent start --focus deployment
+
+# Start in verbose mode
+/agent start --verbose
+
+# Background monitoring
+/agent start --background
+```
+
+**What happens:**
+```
+🤖 Autonomous Admin Agent Starting...
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+Initializing monitoring systems...
+
+✅ Connected to Vercel API (13 projects)
+✅ Connected to Railway API (3 services)
+✅ Connected to Supabase (ekkOS database)
+✅ Connected to MCP server (api.ekkos.dev)
+
+Agent Status: 🟢 Active
+Monitoring: 24/7 continuous
+Check Interval: Every 5 minutes
+Alert Mode: Critical only (will escalate)
+
+───────────────────────────────────────────────────────────────────
+🎯 Monitoring Scope
+
+Deployments:
+  ✅ 13 Vercel projects
+  ✅ 3 Railway services
+  ✅ Pre-deployment checks
+  ✅ Post-deployment monitoring
+
+Memory System:
+  ✅ All 11 layers
+  ✅ Data integrity
+  ✅ Performance metrics
+  ✅ Worker health (PM2)
+
+MCP Server:
+  ✅ 31 tools health
+  ✅ API response times
+  ✅ Error rates
+  ✅ Connection stability
+
+Infrastructure:
+  ✅ Database performance
+  ✅ Storage usage
+  ✅ API rate limits
+  ✅ Service uptime
+
+───────────────────────────────────────────────────────────────────
+🤖 Agent Capabilities
+
+I will automatically:
+  • Run /deploy check before any deployment
+  • Execute /memory-health every hour
+  • Monitor /mcp status every 5 minutes
+  • Optimize database when needed
+  • Alert you for critical issues
+  • Take preventive actions
+  • Learn from past incidents
+
+I will NEVER:
+  • Deploy without confirmation (unless emergency)
+  • Delete data without approval
+  • Override security rules
+  • Take risky actions silently
+
+───────────────────────────────────────────────────────────────────
+📊 Current Status
+
+All systems healthy ✅
+
+Recent Actions (last hour):
+  • 10:15 AM - Ran /memory-health (all healthy)
+  • 10:30 AM - Checked MCP status (31/31 tools OK)
+  • 10:45 AM - Optimized pattern indexes (+12% speed)
+
+Next scheduled checks:
+  • 11:00 AM - Memory health check
+  • 11:05 AM - MCP status check
+  • 11:15 AM - Deployment status review
+
+───────────────────────────────────────────────────────────────────
+✅ Agent Running
+
+Monitoring in background. I'll alert you if I detect issues.
+
+Stop: /agent stop
+Status: /agent status
+Logs: /agent logs
+```
+
+### `/agent status`
+
+Check what the agent is currently doing.
+
+```bash
+/agent status
+
+🤖 Autonomous Admin Agent Status
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+Status: 🟢 Active (running for 3 days, 15 hours)
+
+Current Activity:
+  🔍 Monitoring deployment queue
+  🧠 Analyzing memory layer performance
+  ⏰ Next health check in 2 minutes
+
+───────────────────────────────────────────────────────────────────
+📊 Today's Activity
+
+Checks Performed: 287
+Issues Detected: 3
+Issues Resolved: 2
+Issues Escalated: 1 (requires your attention)
+
+Actions Taken:
+  • Ran /memory-optimize (freed 234 MB)
+  • Restarted stalled Railway worker
+  • Archived 45 stale patterns
+
+Prevented Issues:
+  • Blocked Friday evening deployment (safety rule)
+  • Detected failing tests before deploy
+  • Caught memory leak early (fixed proactively)
+
+───────────────────────────────────────────────────────────────────
+⚠️  Pending Your Attention
+
+1 issue requires human decision:
+
+🔴 apps/admin deployment pending
+   • 2 failing tests detected
+   • Deploy requested by user
+   • Agent blocked (safety rule)
+
+   Options:
+     - Fix tests: Fix the 2 failing tests
+     - Override: /agent approve-deploy apps/admin --reason "..."
+     - Cancel: /agent cancel-deploy apps/admin
+
+───────────────────────────────────────────────────────────────────
+📈 Performance Impact
+
+Since agent started (3 days ago):
+  • Prevented incidents: 5
+  • Reduced manual checks: 287 → 3
+  • Optimization savings: 1.2 GB storage freed
+  • Downtime prevented: ~4 hours estimated
+```
+
+### `/agent stop`
+
+Stop autonomous monitoring (manual mode).
+
+```bash
+/agent stop
+
+🤖 Stopping Autonomous Admin Agent...
+
+✅ Agent stopped.
+
+You're now in manual mode. You can still use admin plugins manually:
+  - /deploy check
+  - /memory-health
+  - /mcp status
+
+Restart: /agent start
+```
+
+### `/agent logs`
+
+View agent's decision log and actions taken.
+
+```bash
+/agent logs
+
+# Last 24 hours
+/agent logs --24h
+
+# Show only actions
+/agent logs --actions-only
+
+# Show only alerts
+/agent logs --alerts-only
+```
+
+**Example:**
+```
+🤖 Agent Activity Log (Last 24 Hours)
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+10:45 AM - 🔍 CHECK
+  Ran: /memory-health
+  Result: All layers healthy
+  Decision: No action needed
+
+10:47 AM - ⚠️  ISSUE DETECTED
+  Problem: Worker judge-queue stalled (6h no activity)
+  Analysis: Database connection timeout
+  Decision: Restart worker
+
+10:48 AM - 🔧 ACTION TAKEN
+  Executed: pm2 restart judge-queue
+  Result: ✅ Worker restarted successfully
+  Outcome: Worker now processing queue
+
+11:00 AM - 🔍 CHECK
+  Ran: /mcp status
+  Result: 31/31 tools healthy, avg 145ms
+  Decision: No action needed
+
+11:15 AM - 🔍 CHECK
+  Ran: /deploy check --all-vercel
+  Result: All deployments healthy
+  Decision: No action needed
+
+11:30 AM - 💡 OPTIMIZATION
+  Detected: Pattern queries slower than baseline (+20%)
+  Analysis: 465 stale patterns reducing index efficiency
+  Decision: Run optimization
+
+11:31 AM - 🔧 ACTION TAKEN
+  Executed: /memory-optimize --layer 4
+  Result: ✅ Freed 234 MB, +15% query speed
+  Outcome: Performance restored
+
+12:00 PM - ⚠️  ISSUE DETECTED
+  Problem: User requesting deploy apps/admin
+  Analysis: 2 tests failing (test/api/users.test.ts)
+  Decision: BLOCK deployment (safety rule)
+
+12:01 PM - 🚨 ALERT SENT
+  Alert: "Deployment blocked - 2 failing tests"
+  Recipient: Admin
+  Action Required: Fix tests or override
+  Status: ⏳ Awaiting admin decision
+
+───────────────────────────────────────────────────────────────────
+📊 Summary
+
+Total Checks: 24
+Issues Detected: 3
+Actions Taken: 2
+Issues Escalated: 1
+```
+
+## Intelligent Decision Making
+
+### Deployment Monitoring
+
+**Agent automatically:**
+1. Runs `/deploy check` before ANY deployment attempt
+2. Blocks deployments with failing tests
+3. Warns about risky deployment times (Friday PM)
+4. Monitors post-deployment for 30 minutes
+5. Auto-rollback if errors spike
+
+**Example scenario:**
+```
+12:00 PM - User runs: vercel deploy --prod
+12:00 PM - Agent intercepts: "Hold on, let me check..."
+12:00 PM - Agent runs: /deploy check apps/memory
+12:01 PM - Agent detects: 2 failing tests
+12:01 PM - Agent blocks: "Deployment blocked - fix tests first"
+12:01 PM - Agent alerts admin: "Deploy requested but blocked"
+```
+
+### Memory Health Monitoring
+
+**Agent automatically:**
+1. Runs `/memory-health` every hour
+2. Detects performance degradation early
+3. Runs `/memory-optimize` when needed
+4. Archives stale patterns proactively
+5. Monitors worker health (PM2)
+
+**Example scenario:**
+```
+10:00 AM - Agent checks: /memory-health
+10:01 AM - Agent detects: L4 queries 20% slower
+10:02 AM - Agent analyzes: 465 stale patterns found
+10:03 AM - Agent decides: Run optimization
+10:04 AM - Agent executes: /memory-optimize --layer 4
+10:05 AM - Agent verifies: Performance restored (+15%)
+10:05 AM - Agent logs: "Proactive optimization complete"
+```
+
+### MCP Server Monitoring
+
+**Agent automatically:**
+1. Checks `/mcp status` every 5 minutes
+2. Tests all 31 tools periodically
+3. Detects slow response times
+4. Restarts server if needed
+5. Alerts for persistent errors
+
+**Example scenario:**
+```
+11:00 AM - Agent checks: /mcp status
+11:01 AM - Agent detects: ekkOS_Search timeout (3 times)
+11:02 AM - Agent analyzes: Database connection pool exhausted
+11:03 AM - Agent decides: Increase pool size
+11:04 AM - Agent executes: Update DATABASE_POOL_SIZE env var
+11:05 AM - Agent restarts: /mcp restart
+11:10 AM - Agent verifies: All tools responding normally
+11:10 AM - Agent logs: "MCP server optimized and restarted"
+```
+
+### Preventive Actions
+
+**Agent takes preventive actions:**
+- Archives stale patterns before they cause slowdowns
+- Optimizes indexes before queries get slow
+- Rotates stale secrets before they expire
+- Increases resources before hitting limits
+- Fixes small issues before they become big
+
+**Example scenario:**
+```
+2:00 PM - Agent analyzes: Memory usage trending up
+2:01 PM - Agent predicts: Will hit 90% in 3 days
+2:02 PM - Agent decides: Archive stale data now (preventive)
+2:03 PM - Agent executes: /memory-optimize --archive-stale
+2:04 PM - Agent result: Freed 1.2 GB, growth delayed
+2:05 PM - Agent logs: "Preventive maintenance complete"
+```
+
+## Alert Levels
+
+### 🟢 INFO (No alert)
+- Routine checks pass
+- Optimizations completed
+- Normal activity
+
+### 🟡 WARNING (Silent)
+- Minor issues detected
+- Performance slightly degraded
+- Agent fixing automatically
+
+### 🟠 ATTENTION (Notify)
+- Issue detected but agent can't fix
+- Requires human decision
+- Non-urgent
+
+### 🔴 CRITICAL (Alert immediately)
+- Production outage
+- Data loss risk
+- Security issue
+- Human intervention required NOW
+
+## Configuration
+
+### `/agent config`
+
+Configure agent behavior.
+
+```bash
+# Set check interval
+/agent config --interval 5m
+
+# Set alert threshold
+/agent config --alert-level warning
+
+# Enable/disable specific monitors
+/agent config --disable deployment-monitoring
+
+# Set working hours
+/agent config --hours "9am-6pm EST"
+```
+
+**Example:**
+```bash
+/agent config
+
+🤖 Agent Configuration
+━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+Monitoring:
+  ✅ Deployment monitoring (every 1m)
+  ✅ Memory health (every 1h)
+  ✅ MCP status (every 5m)
+  ✅ Database performance (every 15m)
+
+Alerts:
+  Level: ATTENTION and above
+  Method: Discord webhook + Email
+  Working Hours: 9 AM - 6 PM EST (escalate after-hours)
+
+Actions:
+  ✅ Auto-optimize (when safe)
+  ✅ Auto-restart workers (when needed)
+  ✅ Auto-archive stale (preventive)
+  ❌ Auto-deploy (requires approval)
+  ❌ Auto-delete data (never)
+
+Safety:
+  ✅ Respect user directives
+  ✅ Block risky deployments
+  ✅ Require approval for destructive actions
+  ✅ Log all decisions
+
+Modify: /agent config --set <key> <value>
+```
+
+## Integration with User Directives
+
+Agent respects ekkOS directives:
+
+```typescript
+// Example directives the agent follows:
+ekkOS_Directive({
+  type: "NEVER",
+  rule: "deploy to production on Friday after 3 PM"
+})
+
+ekkOS_Directive({
+  type: "MUST",
+  rule: "run tests before deploying"
+})
+
+ekkOS_Directive({
+  type: "PREFER",
+  rule: "optimize memory during off-peak hours"
+})
+```
+
+Agent automatically:
+- Reads your directives
+- Applies them to decisions
+- Never violates NEVER/MUST rules
+- Prefers PREFER actions when possible
+
+## MCP Tools Used
+
+The agent uses ALL admin plugins, which in turn use:
+- `ekkOS_Conflict` - Check safety before actions
+- `ekkOS_Search` - Find past incidents
+- `ekkOS_Directive` - Read safety rules
+- `ekkOS_Stats` - Get system metrics
+- `ekkOS_Summary` - Recent activity
+- `ekkOS_Recall` - Past decisions
+- All 31 ekkOS MCP tools
+- Bash commands for system operations
+- Supabase tools for database operations
+
+## Best Practices
+
+### Let the Agent Handle Routine Tasks
+
+**Before agent:**
+```bash
+# You manually run these every day:
+/deploy check --all-vercel
+/memory-health
+/mcp status
+/memory-optimize
+# ... 10+ more commands
+```
+
+**With agent:**
+```bash
+# You just start it once:
+/agent start
+
+# Agent handles everything automatically
+# You only intervene for critical issues
+```
+
+### Review Agent Logs Periodically
+
+```bash
+# Weekly review
+/agent logs --7days --actions-only
+
+# See what agent prevented
+/agent logs --prevented-issues
+```
+
+### Trust but Verify
+
+- Agent takes safe actions automatically
+- Agent alerts you for risky actions
+- You can always override with `/agent approve`
+- All actions are logged for audit
+
+### Emergency Override
+
+```bash
+# If agent blocks something urgent:
+/agent override --reason "Critical security patch"
+
+# Emergency mode (skip all checks):
+/agent emergency-mode --duration 1h
+```
+
+## Troubleshooting
+
+### Agent Not Starting
+
+**Problem:** `/agent start` fails
+**Check:** ekkOS MCP connection
+**Fix:** Authenticate ekkos-connect extension
+
+### Agent Taking Wrong Actions
+
+**Problem:** Agent optimizing at wrong times
+**Fix:** Adjust config: `/agent config --optimize-hours "2am-4am"`
+
+### Too Many Alerts
+
+**Problem:** Getting alerted for minor issues
+**Fix:** Raise threshold: `/agent config --alert-level critical`
+
+### Agent Missed an Issue
+
+**Problem:** Issue occurred but agent didn't detect
+**Fix:** Check logs: `/agent logs --show-missed`
+**Report:** Help improve agent by reporting missed issues
+
+---
+
+## Summary
+
+The Autonomous Admin Agent is your AI DevOps engineer that:
+
+✅ **Monitors** - 24/7 infrastructure monitoring
+✅ **Detects** - Issues before they become problems
+✅ **Decides** - Intelligently chooses actions
+✅ **Acts** - Safe automatic fixes
+✅ **Alerts** - Only when you're needed
+✅ **Learns** - Gets smarter over time
+
+**Start it once, forget routine maintenance.**
+
+```bash
+/agent start
+```
+
+---
+
+**Monitor intelligently. Act automatically. Sleep peacefully.** 🤖

package/templates/claude-plugins-admin/backend-agent/.claude-plugin/plugin.json

@@ -0,0 +1,8 @@
+{
+  "name": "ekkos-admin-backend",
+  "description": "ADMIN ONLY: AI Backend Developer specialized in API development, database design, Supabase, PostgreSQL, and server-side logic.",
+  "author": {
+    "name": "ekkOS Internal",
+    "email": "dev@ekkos.dev"
+  }
+}